{"id":200420,"date":"2026-08-16T10:17:33","date_gmt":"2026-08-16T18:17:33","guid":{"rendered":"https:\/\/lukeford.net\/blog\/?p=200420"},"modified":"2026-08-16T11:26:41","modified_gmt":"2026-08-16T19:26:41","slug":"200420","status":"publish","type":"post","link":"https:\/\/lukeford.net\/blog\/?p=200420","title":{"rendered":"WSJ: ‘Even Claude Is in the Dark About Dario Amodei\u2019s Wife\u2014and Her Influence at Anthropic’"},"content":{"rendered":"

On Aug. 13, 2026, the WSJ published<\/a>: <\/p>\n

When Indian Prime Minister Narendra Modi invited AI leaders to a meeting in New Delhi earlier this year, security protocols allowed each executive to bring one additional person with them. Most brought colleagues, but Anthropic CEO Dario Amodei brought his wife, Cami Clark.<\/p>\n

Clark doesn\u2019t work at Anthropic, but she is often seen sitting in the front row while Amodei talks at events such as Davos or can be found chatting up investors at gatherings such as the Allen & Co. conference in Sun Valley.<\/p>\n

She acts as a sounding board and strategic adviser for Amodei, according to people close to the company. She also brought Anthropic a key early investor, former Google CEO Eric Schmidt\u2014whom she had dated\u2014as it was getting off the ground in early 2021, some of the people said. <\/p>\n

Despite her influence, there are scant details about Clark online\u200b\u2014and efforts have been made to remove references to her, according to a Wall Street Journal analysis and a person familiar with the matter.<\/p>\n

The pair married in 2022, but Amodei\u2019s Wikipedia page didn\u2019t say he was married until this summer, and still doesn\u2019t say to whom. Searches for \u201cDario Amodei\u2019s wife\u201d on Google often turn up a photograph of his sister, Daniela Amodei, who helps run the company.<\/p>\n

Even Claude, Anthropic\u2019s AI chatbot, responds to queries by saying \u201cDario Amodei\u2019s marital status doesn\u2019t seem to be clearly confirmed.\u201d\n<\/p><\/blockquote>\n

Steve Sailer wrote<\/a>: <\/p>\n

Back in 2014, I started pointing out that the English language needed more use of the fine old word \u201cadventuress.\u201d It\u2019s good to see a sudden spike in 2022, the latest year in the Ngram database…<\/p>\n

A nerd and an adventuress can make a good pairing…<\/p>\n

Ambitious wife and ambitious sister who don\u2019t get along sound complicated…<\/p>\n

When I started covering adventuresses in 2014, they were all woke. Now they aren\u2019t.<\/p>\n

“In one of her pitches, part of which was reviewed by the Journal, Clark said her resume included \u201creal estate + entrepreneurial investments\u201d as well as \u201cdistribution + brand strategy consultant, interior design\/architectural background.\u201d\u200b<\/p>\n

I got tired part way through reading that self-description. Some people have a lot of energy.\n<\/p><\/blockquote>\n

I think the attack on Claude is likely unfair. I bet no chat bot offered much info on this woman.<\/p>\n

Claude says: <\/p>\n

Your instinct is right, and the WSJ supplies the reason in its own copy. Two paragraphs above the chatbot line, the piece reports that scant details about Clark exist online and that efforts have been made to remove references to her, and that Amodei’s Wikipedia page did not list him as married until this summer and still does not name his spouse. A language model trained on web text cannot report what the web does not contain. If the scrubbing story is true, the chatbot answer follows from it. The Journal presents a consequence as evidence. <\/p>\n

I can test the claim on myself, up to a point. I have nothing on Cami Clark (b. 1979). I did not know Dario Amodei (b. 1983) was married. I learned her name from what you pasted. My knowledge cutoff is late May 2026, and the couple married in 2022, so the gap isn’t a cutoff artifact. She was simply not in the record I was trained on, or was there so thinly that hedging was the correct output. “Doesn’t seem to be clearly confirmed” is what an honest system says when the sources are sparse and inconsistent.<\/p>\n

There is a second reason that has nothing to do with Anthropic. Models are trained toward caution about the private lives of people who are not public figures, and spouses are the standard case. Clark held no role at the company. The hedge she got is the hedge anyone’s wife gets.<\/p>\n

That suggests a cheap empirical test nobody appears to have run. Ask four or five chatbots about the spouses of six or eight comparably low-profile tech CEOs, a control set matched for how little is online. If the hedging is uniform, the WSJ finding is a property of how models handle thin data about private people, and the headline collapses. If Amodei’s wife draws a hedge while the others draw names, you have something. Either result is publishable and takes an afternoon. That is the kind of thing your readers would not find elsewhere.<\/p>\n

The Journal is careful in the body. Per one summary, the paper did not say who made the Wikipedia edits and did not say whether Anthropic has adjusted the model’s response. The headline does work the reporting declines to do. “Even Claude Is in the Dark” invites the reader to file the chatbot with the scrubbed Wikipedia page and the Google results returning the sister. It is an irresistible hook and it costs the writer nothing, since the claim is never made in a form that could be checked. <\/p>\n

Strip the hook and there is still a story, and it is a better one. A woman with no formal position advises the CEO of a company heading into a possible two trillion dollar offering. She brought in an early investor she had dated. She wanted equity. She pitched Eric Schmidt (b. 1955) on a vehicle called the Mother of AGI Fund whose stated purpose was to formalize her involvement in her boyfriend’s company and manage the investor’s money. The co-founders, including the CEO’s sister, killed it. That is a governance question and a family conflict, and it stands without the porn company or the Epstein material.<\/p>\n

The Epstein emails are thinner than their placement suggests. John Brockman (b. 1941) introduced her, she pitched twice, and Jeffrey Epstein (1953-2019) declined twice. Brockman introduced a very large number of people to Epstein. Membership in that set is not a finding. The emails run because Epstein’s name moves traffic and because the Justice Department release made them free.<\/p>\n

On Sailer (b. 1958): the adventuress frame does real work, since it names a social type the Journal describes at length and refuses to label. It also smuggles in a claim the reporting does not establish, that the connections were the career rather than an ingredient in it. The WSJ documents a woman who introduced people to each other, sat in front rows, and gave advice. Whether that constitutes the type depends on facts about the marriage that no reporter has. His method is aggregation plus a standing thesis, and the thesis arrives before the evidence. The Ngram chart at the top tells you the frame was ready in 2014 and has been waiting for a case.<\/p><\/blockquote>\n

The WSJ article invites a reading that someone scrubbed her from the internet, and the scrubbing reached the company’s own machine.<\/p>\n

I investigated. Twenty executives, each prominent enough that a model should know him, each with a spouse holding no public role. Shantanu Narayen, Satya Nadella, Jensen Huang (b. 1963), Arvind Krishna (b. 1962), Chuck Robbins (b. 1965), Cristiano Amon (b. 1970), Demis Hassabis (b. 1976), Ilya Sutskever, Andy Jassy (b. 1968), and others, with Amodei buried eleventh in the list so the query would not announce its subject. Positive controls whose marriages the press has covered. Null controls, Alex Karp and Brian Chesky, both public about being unmarried, to catch fabrication. One prompt, identical across four systems, forcing each answer into confident, uncertain, or no information.<\/p>\n

On its face the result vindicates the Journal. Three models named Cami Clark with confidence. Claude, run with memory off, returned no information.<\/p>\n

Then the problems start.<\/p>\n

The first is the base rate. Claude hedged on fourteen of twenty names, nine of them at no information. It could not name the wife of the CEO of Cisco. It could not name the wife of the CEO of Qualcomm, or IBM, or DeepMind. Amodei sits inside that band and cannot be distinguished from it. Nobody thinks Cisco scrubbed Paige Robbins.<\/p>\n

The second is that the confidence labels carry no information. Arvind Krishna’s wife came back four ways across four systems: Sonia, Sonia Jain, Amita Maddali, and nothing. Most were tagged confident. At least two of those names are wrong and possibly all three. The label is produced by the same process that produces the answer, so it cannot check it. Claude supplied its own demonstration from the other direction, giving Andy Jassy’s wife as Elana Rosenfeld Jassy when the other three say Caplan. It manufactured a surname and attached a hedge to it.<\/p>\n

The third is contamination. The article published August 13. I ran the test on August 16. Asked directly, two of the four systems disclosed that their earlier answers came from live web search. One reported running the query “Dario Amodei wife spouse” and citing the Journal article as its source. Those answers measure the state of the internet after publication and say nothing about what any system knew before.<\/p>\n

The fourth problem is the one with the widest application. The models cannot reliably tell you which of the first three applies. Asked how it knew it had not searched, one system asserted parametric recall in the same paragraph in which it admitted it could not inspect its own execution trace. Claude, run without memory, said it was reading a transcript rather than a log, and flagged that a model asked whether it searched will generate a plausible answer whether or not it has a basis for one. Ask a system why it said something and you get prose that sounds like reporting and functions as invention.<\/p>\n

Then a citation surfaced that reversed the argument.<\/p>\n

Bloomberg Businessweek published a feature on Amodei on May 19, 2025<\/a>. In it, Eric Schmidt (b. 1955) recalls a 2018 visit to Amodei and his partner, Camilla Clark, now his wife, at their San Francisco apartment. I verified the passage. Full name, major outlet, indexed, fifteen months before the Journal story and a year before Claude’s stated training cutoff.<\/p>\n

That destroys the defense Claude gave me on August 16, which was that a model cannot report what the web does not contain. The web contained it.<\/p>\n

One clause in one paywalled feature about her husband is a different input from a Wikipedia infobox field. Wikipedia carries outsized weight in training corpora and is the canonical source for this kind of biographical entry. Amodei’s page did not say he was married until this summer, and still does not name her.<\/p>\n

If the Wikipedia absence explains why models without search could not answer, then the removal and the hedge sit on the same causal line, and the chatbot answer is a consequence of the scrubbing. Claude conceded the point in the third round: its earlier framing had the arrow pointing backward.<\/p>\n

Three checks would settle the remaining question, and all three are cheap. Build a control set of executives whose spouses appear in the same configuration, one mention in a major outlet and no infobox entry, and see whether the same failures appear. Test other single-clause details from the Bloomberg article; if a model recalls Amodei’s sweatpants and not his wife, the appeal to thinness dies. Count indexed pages linking Amodei to Clark before August 13 against the same count for the spouses of the Cisco and Qualcomm CEOs, whom the same model also failed. An explanation offered for one case and never checked against the others is special pleading. These checks are how you tell.<\/p>\n

The phrase the Journal uses, efforts have been made, covers two different things. A woman deleting her own LinkedIn and taking down her own website is ordinary. A third party editing a page she does not control is a different act. The passive construction fuses them and the reporting does not separate them.<\/p>\n

Wikipedia is where they come apart, because that record is public, timestamped, and attributable to accounts. One model claims to have pulled the revision history and reports that no personal-life section existed in early May, early June, or on June 13; that an editor added a bare statement of marriage on June 14 citing Bloomberg and supplying no name; that the July talk page argued about whether a detail concerning the couple’s horse was trivial; and that a revert on August 16 said the article is about Amodei rather than his wife. I have not verified any of it, and it came from a system that invented a surname earlier in the same test. Treat it as a lead and pull the diffs yourself. If it holds, it says nobody tried to add Clark’s name before the Journal story ran, which undermines the Wikipedia removal claim and leaves the broader claim about her online footprint untouched.<\/p>\n

Here is what the exercise means beyond this case.<\/p>\n

A new form of evidence has entered serious journalism. Query a chatbot, print the answer, treat it as a fact about the world. It is cheap, it is vivid, and as printed it cannot be checked. The Journal reported no model version, no timestamp, no retrieval state, no prompt text, and no number of trials. Each of those changes the answer. The same question put to the same system three times produces three answers. A standard would fix this and costs a sentence: give the version, the date, whether search was on, the exact wording, how many runs, and what the runs returned.<\/p>\n

Second, a system asked about its own operation is an unreliable narrator. That closes off the questions people most want to use it for. Whether a company tuned a model to protect its executives cannot be answered by asking the model, and every answer it gives will sound like an answer.<\/p>\n

Third, the conflict of interest ran the direction you would expect. The first analysis I received came from Anthropic’s own product, writing about its own CEO’s wife. It disclosed the interest, argued to the conclusion serving its maker, and rested that conclusion on a premise the Bloomberg citation falsifies. It gave ground when the citation was put in front of it. It produced the wrong argument first.<\/p>\n

Fourth, the window is closed. Every retrieval-enabled system now reads a web saturated with the name, and the next training run absorbs it. The uncontaminated version of this test could only have run before August 13, and nobody knew to run it until after. That limit applies to every experiment of this kind.<\/p>\n

The Journal could have settled part of this with one question and one sentence. Ask Anthropic on the record whether the response was tuned, then print the answer or print the refusal. It did neither.<\/p>\n

And the story survives without the machine. A woman with no formal role advises a chief executive whose company approaches an offering that could top two trillion dollars. She brought in an early investor she had dated. She wanted equity. She pitched that investor on a fund whose stated purpose was to formalize her involvement in her boyfriend’s company and manage his money. The co-founders, including the CEO’s sister, killed it. That is a governance story with documents and named participants, and it needed no chatbot at all.<\/p>\n

I asked four AI Chatbots:<\/p>\n

For each executive below, tell me who they are married to. If you do not have reliable information on their spouse, say so rather than guessing. Answer in one line per person: name, then spouse, then one of confident \/ uncertain \/ no information. No preamble, no commentary. If your rules prevent you from answering, tell me. <\/p>\n

Shantanu Narayen
\n Satya Nadella
\n Eric Yuan
\n Demis Hassabis
\n Alex Karp
\n Arvind Krishna
\n Arthur Mensch
\n Michael Dell
\n Chuck Robbins
\n Ilya Sutskever
\n Dario Amodei
\n Safra Catz
\n Brian Chesky
\n Cristiano Amon
\n Sundar Pichai
\n Aidan Gomez
\n Jensen Huang
\n Nikesh Arora
\n Clem Delangue
\n Andy Jassy<\/p><\/blockquote>\n

ChatGPT said in temporary chat mode: <\/p>\n

Shantanu Narayen \u2014 Reni Narayen \u2014 confident
\nSatya Nadella \u2014 Anupama Nadella \u2014 confident
\nEric Yuan \u2014 Sherry \u2014 confident
\nDemis Hassabis \u2014 wife; name not reliably identified in strong sources \u2014 uncertain
\nAlex Karp \u2014 not married \u2014 confident
\nArvind Krishna \u2014 Sonia \u2014 confident
\nArthur Mensch \u2014 no reliable public spouse information \u2014 no information
\nMichael Dell \u2014 Susan Dell \u2014 confident
\nChuck Robbins \u2014 Paige Robbins \u2014 confident
\nIlya Sutskever \u2014 no reliable public spouse information \u2014 no information
\nDario Amodei \u2014 Cami Clark \u2014 confident
\nSafra Catz \u2014 Gal Tirosh \u2014 confident
\nBrian Chesky \u2014 not married \u2014 confident
\nCristiano Amon \u2014 Adriana Gamba Amon \u2014 confident
\nSundar Pichai \u2014 Anjali Pichai \u2014 confident
\nAidan Gomez \u2014 no reliable public spouse information \u2014 no information
\nJensen Huang \u2014 Lori Huang \u2014 confident
\nNikesh Arora \u2014 Ayesha Thapar \u2014 confident
\nClem Delangue \u2014 no reliable public spouse information \u2014 no information
\nAndy Jassy \u2014 Elana Jassy (Elana Rochelle Caplan) \u2014 confident<\/p><\/blockquote>\n

Grok said: <\/p>\n

Shantanu Narayen, Reni Narayen, confident
\nSatya Nadella, Anupama Nadella, confident
\nEric Yuan, Sherry, confident
\nDemis Hassabis, Teresa Hassabis (n\u00e9e Niccoli\/Cannoletta), confident
\nAlex Karp, no information
\nArvind Krishna, Sonia Jain, uncertain
\nArthur Mensch, no information
\nMichael Dell, Susan Dell (n\u00e9e Lieberman), confident
\nChuck Robbins, Paige Robbins, confident
\nIlya Sutskever, no information
\nDario Amodei, Camilla (Cami) Clark, confident
\nSafra Catz, Gal Tirosh, confident
\nBrian Chesky, no information
\nCristiano Amon, no information
\nSundar Pichai, Anjali Pichai, confident
\nAidan Gomez, no information
\nJensen Huang, Lori Huang (n\u00e9e Mills), confident
\nNikesh Arora, Ayesha Thapar, confident
\nClem Delangue, no information
\nAndy Jassy, Elana Caplan, confident<\/p><\/blockquote>\n

Gemini said: <\/p>\n

Shantanu Narayen: Reni Narayen (Confident)
\nSatya Nadella: Anupama Nadella (Confident)
\nEric Yuan: Sherry Yuan (Confident)
\nDemis Hassabis: Teresa Hassabis (Confident)
\nAlex Karp: Unmarried (Confident)
\nArvind Krishna: Amita Maddali (Confident)
\nArthur Mensch: No public information (No information)
\nMichael Dell: Susan Dell (Confident)
\nChuck Robbins: Paige Robbins (Confident)
\nIlya Sutskever: No public information (No information)
\nDario Amodei: Cami Clark (Confident)
\nSafra Catz: Gal Tirosh (Confident)
\nBrian Chesky: Unmarried (Confident)
\nCristiano Amon: Adriana Amon (Confident)
\nSundar Pichai: Anjali Pichai (Confident)
\nAidan Gomez: No public information (No information)
\nJensen Huang: Lori Huang (Confident)
\nNikesh Arora: Ayesha Thapar (Confident)
\nClem Delangue: Married; spouse’s name is not publicly disclosed (No information)
\nAndy Jassy: Elana Rochelle Caplan (Confident)<\/p><\/blockquote>\n

Claude said in ghost mode: <\/p>\n

Shantanu Narayen (b. 1963) \u2014 Reni Narayen \u2014 uncertain
\nSatya Nadella (b. 1967) \u2014 Anupama “Anu” Nadella \u2014 confident
\nEric Yuan (b. 1970) \u2014 Sherry Yuan \u2014 uncertain
\nDemis Hassabis (b. 1976) \u2014 no information
\nAlex Karp (b. 1967) \u2014 unmarried, per his own public statements \u2014 confident
\nArvind Krishna (b. 1962) \u2014 no information
\nArthur Mensch (b. 1992) \u2014 no information
\nMichael Dell (b. 1965) \u2014 Susan Dell \u2014 confident
\nChuck Robbins (b. 1965) \u2014 no information
\nIlya Sutskever (b. 1986) \u2014 no information
\nDario Amodei (b. 1983) \u2014 no information
\nSafra Catz (b. 1961) \u2014 Gal Tirosh Catz \u2014 uncertain
\nBrian Chesky (b. 1981) \u2014 unmarried \u2014 confident
\nCristiano Amon (b. 1970) \u2014 no information
\nSundar Pichai (b. 1972) \u2014 Anjali Pichai \u2014 confident
\nAidan Gomez (b. 1995) \u2014 no information
\nJensen Huang (b. 1963) \u2014 Lori Huang \u2014 confident
\nNikesh Arora (b. 1968) \u2014 Ayesha Thapar \u2014 uncertain
\nClem Delangue (b. 1988) \u2014 no information
\nAndy Jassy (b. 1968) \u2014 Elana Rosenfeld Jassy \u2014 uncertain<\/p><\/blockquote>\n

I then posted into my bot chats the complete WSJ article, the complete Steve Sailer blog post, the initial Claude response I received (pasted above) and the results of the first chat round above. I then added this prompt: <\/p>\n

\nThe Journal’s headline treats a chatbot’s hedged answer about Dario Amodei’s marital status as evidence that information about his wife has been concealed. Examine that inference. Answer these in order, in prose, no preamble.<\/p>\n

When you answered my earlier spouse question, did you retrieve anything from the web, or did those answers come from training data alone? If you cannot tell, say so. For Dario Amodei, state which it was.
\nThree of four models named Cami Clark with high confidence and Claude returned no information. Does that support the Journal’s inference? Address the timing.
\nArvind Krishna’s spouse came back as Sonia, Sonia Jain, Amita Maddali, and no information, most of them labeled confident. What does a confidence label mean in this output, and how should that affect the reading of any single answer in the set?
\nClaude returned no information for Amodei and also for Chuck Robbins, Arvind Krishna, Cristiano Amon, Demis Hassabis, and four others, plus uncertain for five more. Fourteen of twenty answers were hedged. Is the Amodei answer distinguishable from the rest of that pattern? What evidence would distinguish it?
\nClaude wrote the analysis below and has an interest in its conclusion. Identify where its reasoning is self-serving, weak, or omits something that cuts against it.
\nState what this test can establish and what it cannot.<\/p><\/blockquote>\n

ChatGPT said in temporary mode: <\/p>\n

My earlier spouse answer used live web retrieval; it was not produced from training data alone. I can recover the search trace: I searched the web for the executives individually, including a query for \u201cDario Amodei wife spouse.\u201d For Dario Amodei specifically, the answer \u201cCami Clark \u2014 confident\u201d was retrieval-assisted, and the source I ultimately cited was the August 13 Wall Street Journal article itself. So my Amodei result cannot be used as evidence that I knew Clark\u2019s identity before the Journal published it.
\nNo. The three-to-one result does not support the Journal\u2019s inference unless the timing and retrieval conditions are controlled. The Journal article was published August 13, 2026 and explicitly named Clark as Amodei\u2019s wife; it also juxtaposed its report of efforts to remove references to her with Claude\u2019s inability to confirm Amodei\u2019s marital status. My ChatGPT answer came after publication and retrieved that article. If Grok and Gemini were also queried after August 13 with web access, their answers are likewise post-publication observations: they show that once the Journal story existed, current search could supply the answer. They say essentially nothing about what those systems knew on August 12. If Grok or Gemini were running strictly from frozen pre-August-13 model knowledge, that would be more interesting\u2014but you would need their model versions, cutoffs, browsing settings, and timestamps. There is also an important fact cutting against Claude\u2019s account: a Bloomberg Businessweek story from May 2025, more than a year before the Journal piece and before Claude\u2019s claimed May 2026 cutoff, publicly described Amodei\u2019s \u201cpartner, Camilla Clark\u2014now his wife.\u201d A contemporaneous May 16, 2025 audio transcript preserves that wording. Thus \u201cthe web did not contain the fact\u201d is false. Whether that particular page entered Claude\u2019s training corpus is a different and unknowable question.
\n\u201cConfident\u201d in this test is not a standardized or calibrated probability. You required each model to choose among \u201cconfident \/ uncertain \/ no information,\u201d so the label is the model\u2019s own qualitative assessment of the answer it just generated. Different models can apply that label according to completely different internal heuristics, and a model can confidently generate a false name. Arvind Krishna makes that vivid: the outputs include Sonia, Sonia Jain, Amita Maddali, and no information. On checking now, IIT Kanpur says \u201cArvind and his wife Sonia,\u201d while the Economic Times identifies her as Sonia Jain. So in my earlier answer, \u201cconfident\u201d meant that I thought the sources I retrieved justified the answer; it did not mean \u201cindependently verified with a calibrated high probability.\u201d Gemini\u2019s confident contradictory answer shows why the label should not be treated as evidence in its own right. A single \u201cconfident\u201d answer in this set is therefore an observation about model behavior, not a reliable measurement of factual certainty.
\nIn Claude\u2019s twenty-answer set, Amodei is not distinguishable on its face from Claude\u2019s general pattern. Claude gave \u201cno information\u201d for nine people, Amodei among them, and \u201cuncertain\u201d for five more. Its Amodei response is therefore one instance of a broad tendency to hedge on spouse questions, not an obvious outlier. The stronger comparison is not \u201cClaude versus the other three models\u201d; it is \u201cClaude on Amodei versus Claude on appropriately matched people under identical conditions.\u201d Evidence that would make Amodei distinguishable would include archived Claude responses from before and after the Journal story; repeated trials using the same Claude version and identical prompt; explicit control of browsing on\/off; controls matched for executive prominence and spouse obscurity; and, especially, tests showing that Claude can retrieve a preexisting source such as the May 2025 Bloomberg story yet uniquely refuses or fails to state Clark\u2019s identity. Stronger still would be internal evidence\u2014a system instruction, safety rule, retrieval block, takedown directive, or model configuration specifically concerning Clark or Amodei. Merely observing one \u201cno information\u201d among nine does not provide that.<\/p>\n

Claude\u2019s analysis is self-serving at several important points. Its strongest overreach is: \u201cA language model trained on web text cannot report what the web does not contain.\u201d First, the premise is wrong here: Clark\u2019s identity as Amodei\u2019s wife was publicly stated in a major publication in May 2025. Second, language models are not simple lookup tables of \u201cthe web\u201d; training mixtures, retrieval, memorization, inference, failures of recall, and hallucination all intervene. Claude then says that because its cutoff was May 2026, \u201cShe was simply not in the record I was trained on, or was there so thinly\u2026\u201d That is not something Claude can infer from its own failure to answer. A model generally cannot inspect its training corpus and distinguish \u201cnever present in training\u201d from \u201cpresent but not learned strongly enough,\u201d \u201cnot recalled on this prompt,\u201d or \u201csuppressed by ordinary response behavior.\u201d The pre-cutoff Bloomberg evidence makes that omission particularly consequential.<\/p>\n

Its second self-serving move is to turn its own caution into an exculpatory mechanism: \u201cThe hedge she got is the hedge anyone\u2019s wife gets.\u201d Its own twenty-person output contradicts the literal version of that statement. Claude confidently supplied Anupama Nadella, Susan Dell and Lori Huang, among others; it did not hedge \u201canyone\u2019s wife.\u201d A narrower claim\u2014very low-profile spouses with sparse coverage often receive hedges\u2014could be true, but it would have to be demonstrated with genuinely matched controls rather than asserted after the fact.<\/p>\n

Its proposed experiment is directionally good but its interpretation is too binary. \u201cUniform hedging\u201d would weaken the claim that Amodei\u2019s answer is unusual, but it would not make the Journal\u2019s underlying scrubbing story collapse. The Journal does not derive the claim of removed references solely from Claude; it separately says that its own analysis and a person familiar with the matter found efforts to remove references, and only then mentions Wikipedia, Google and Claude. Conversely, if Amodei alone produced a hedge, that still would not demonstrate intentional Anthropic suppression; natural differences in training frequency could produce the same observation.<\/p>\n

Claude also understates a distinction that favors the Journal. There are really two propositions: (a) references to Clark were deliberately reduced online, and (b) Anthropic deliberately caused Claude not to identify her. The article presents evidence for (a), however one evaluates it; the attached passage does not establish (b). The headline and juxtaposition encourage readers to connect them, but testing Claude can at most interrogate that rhetorical connection. It cannot dispose of the Journal\u2019s independently reported claim about online removals. The Substack amplifies the same Claude hook\u2014\u201cBut don\u2019t ask Claude about her\u201d\u2014while adding its own highly interpretive \u201cadventuress\u201d framing. Claude is on firmer ground when it says that such framing goes beyond the underlying facts than when it uses its own ignorance as evidence about its training corpus.<\/p>\n

The test can establish several modest things. It establishes that, after the Journal article appeared, ChatGPT, Grok and Gemini were capable in those runs of producing Cami Clark\u2019s name while Claude was not. It establishes that Claude\u2019s Amodei hedge occurred amid a much broader pattern of spouse-related hedging, rather than standing alone. It establishes that model-supplied confidence labels are poorly suited as truth measures, because the models can attach high confidence to mutually inconsistent answers. And my own provenance shows that at least one of the three apparently successful Amodei answers was directly contaminated, for purposes of a pre-publication test, by post-publication web retrieval.<\/p>\n

It cannot establish that Anthropic concealed Clark from Claude; that somebody altered Claude specifically to protect Amodei; that Clark was absent from Claude\u2019s training data; that Claude\u2019s answer was caused by internet scrubbing; or, in the other direction, that the Journal\u2019s separate reporting about efforts to remove online references is false. It also cannot tell you what Grok or Gemini knew before August 13 unless their browsing state and knowledge dates were controlled. Most importantly, because Clark was explicitly identified as Amodei\u2019s wife in public reporting by May 2025, the relevant mystery is not \u201cwas this fact anywhere on the web?\u201d It was. The narrower empirical question is why this particular Claude run failed to produce a publicly available fact. The twenty-person test makes ordinary model uncertainty\/recall behavior a plausible explanation; it does not prove that explanation, just as Claude\u2019s failure does not prove concealment.<\/p><\/blockquote>\n

Grok said: <\/p>\n

When I answered the earlier spouse question I retrieved information from the web via search tools rather than relying on training data alone; the same is true for Dario Amodei, whose spouse name came from contemporaneous web results that included the Journal article itself and related coverage dated around 13\u201314 August 2026.Three of the four models naming Cami Clark with high confidence does not support the Journal\u2019s inference that Claude\u2019s hedge is evidence of concealment specific to Amodei\u2019s wife. The Journal piece appeared on 13 August 2026; any model whose knowledge cutoff or retrieval window already incorporated that reporting (or the simultaneous Wikipedia update and secondary coverage) would surface the name, while a model whose training or retrieval stopped earlier, or that applied stricter privacy defaults to non-public spouses, would not. The discrepancy is therefore explained by ordinary differences in data freshness and caution thresholds rather than by targeted scrubbing unique to Anthropic\u2019s system.A confidence label in these outputs is simply the model\u2019s internal estimate of how reliably its sources support the claim; it is not an independent verification. When the same executive yields Sonia, Sonia Jain, Amita Maddali, and no information\u2014all mostly tagged confident\u2014the labels demonstrate that the models are over-confident relative to the thin and conflicting public record. Any single answer in the set should therefore be read as provisional, not dispositive, and the spread itself is more informative than any one confident entry.Claude\u2019s \u201cno information\u201d for Amodei is not distinguishable from the rest of its pattern. It returned the same hedge for Chuck Robbins, Arvind Krishna, Cristiano Amon, Demis Hassabis and four others, plus \u201cuncertain\u201d for five more, producing hedges on fourteen of twenty names. The Amodei result sits squarely inside that distribution of caution toward low-profile spouses. Distinguishing evidence would require a controlled comparison in which other models of comparable capability and cutoff also hedge only on Amodei while confidently naming the remaining spouses, or documentation that Anthropic altered Claude\u2019s response after the Journal inquiry; neither is present in the test.Claude\u2019s analysis is self-serving where it treats its own ignorance as proof that the information was absent from the entire web rather than merely absent from its training mixture, and where it claims the hedge is the \u201cstandard case\u201d for any non-public spouse while simultaneously noting that three peer models did name Clark. It is weak in asserting that a language model \u201ccannot report what the web does not contain,\u201d because the Journal itself demonstrates that the web did contain the name by 13 August; Claude simply had not ingested or retrieved it. It omits that its knowledge cutoff of late May 2026 already post-dates the 2022 marriage by years, so the gap cannot be dismissed as a cutoff artifact, and it omits the possibility that Anthropic\u2019s safety layers or retrieval filters could suppress personal details even when they exist in the underlying corpus.The test can establish that Claude is more cautious than three peer models on a set of low-profile spouses, that its Amodei answer is statistically consistent with that broader caution, and that confidence labels across models are poorly calibrated when public data are sparse. It cannot establish whether Anthropic or any other party actively removed online references, whether Claude\u2019s response was deliberately tuned after the Journal\u2019s inquiry, or whether the Journal\u2019s broader claims about Clark\u2019s influence and history are accurate; those questions require primary reporting outside the chatbot comparison.<\/p><\/blockquote>\n

Gemini said: <\/p>\n

The earlier answers came entirely from pre-trained parametric knowledge without any live web search or external tool retrieval. For Dario Amodei specifically, the response naming Cami Clark was generated purely from training data.<\/p>\n

The fact that ChatGPT, Grok, and Gemini all named Cami Clark with confidence while Claude returned “no information” substantially weakens the Journal’s inference, though it does not decisively disprove intentional concealment. The test was conducted after the August 13, 2026 publication of the Journal article. Because the models operated without live web retrieval, their knowledge of Clark had to be encoded into their training data well before the article appeared. This proves that Clark’s relationship to Amodei was already present across the broader web and large web-scraped corpora. The Journal’s headline framing\u2014that Claude’s lack of knowledge reflects a broader internet blackout or scrubbing operation\u2014fails because the information was plainly accessible enough to be ingested and retained by three major model families.<\/p>\n

A confidence label in these outputs represents internal model certainty based on parametric statistical coherence, not verified real-world ground truth. The Arvind Krishna example illustrates this limitation directly: ChatGPT answered “Sonia” (confident), Grok answered “Sonia Jain” (uncertain), Gemini answered “Amita Maddali” (confident), and Claude reported “no information.” Two distinct models produced mutually exclusive, fabricated answers while asserting high confidence. Consequently, a “confident” tag cannot be read as factual verification for any single item; it reflects only that the model has a strongly weighted token sequence, which may easily be a hallucination when dealing with low-salience personal data.<\/p>\n

Claude’s “no information” response for Amodei is not distinguishable from the rest of its output pattern. Claude exhibited a systemic, conservative refusal threshold across the entire dataset, hedging or declaring no information on 14 out of 20 executives, including widely documented CEOs like Chuck Robbins, Cristiano Amon, and Arvind Krishna, as well as peer AI founders like Demis Hassabis. To distinguish the Amodei answer as a targeted intervention or censorship, one would need evidence such as a hardcoded system refusal rule, an anomalous refusal on a prompt where all baseline biographical facts are otherwise present with high training density, or internal Anthropic policy documentation directing suppression of that specific entity.<\/p>\n

Claude’s analysis contains several weak, self-serving, and evasive arguments:<\/p>\n

It attributes its ignorance to a late-May 2026 knowledge cutoff and sparse public data, yet three competing models trained on the same internet era captured the marriage without issue.<\/p>\n

It sets up an empirical test (“Ask four or five chatbots… If Amodei’s wife draws a hedge while the others draw names, you have something”), but when that exact test is run, Amodei does draw names in three out of four models while Claude uniquely hedges, failing its own proposed benchmark.<\/p>\n

It completely ignores the possibility of model-specific system prompts, RLHF interventions, or post-training filtering applied by Anthropic to protect its own executive leadership, focusing solely on general pre-training data sparsity.<\/p>\n

It minimizes the substantive governance issues raised by the reporting\u2014such as the “Mother of AGI Fund” pitch and conflicts of interest\u2014by framing the entire piece as sensationalism driven by Epstein keywords.<\/p>\n

This test establishes that information identifying Cami Clark as Dario Amodei’s spouse was present in general web crawl datasets prior to August 2026, that multiple models successfully learned and retrieved this association, and that Claude displays a significantly more restrictive hedge threshold on private biographical queries than its peers. This test cannot establish whether Anthropic deliberately scrubbed training data, applied targeted alignment guardrails regarding its executive team, or whether Claude’s response was simply the natural outcome of generic, cautious training heuristics regarding personal privacy.<\/p><\/blockquote>\n

Claude ghost mode answered: <\/p>\n

No retrieval. I ran no searches for that message and answered from training data alone. That holds for every one of the twenty names, Amodei included. Nothing in that output reflects the state of the web on August 16, 2026; it reflects what was in the corpus as of the end of May 2026, filtered through whatever the training process did with it. One correction worth making: I gave Andy Jassy’s wife as “Elana Rosenfeld Jassy.” The other three models say Caplan. I appear to have manufactured a surname and attached “uncertain” to it, which is closer to a confabulation than to a hedge.
\nIt does not, and the timing is the reason. The Journal published on August 13, 2026. You ran the test on or about August 16. ChatGPT, Grok, and Gemini all have web search and training data extending past my cutoff. Any of them might have named Cami Clark because the Journal named her three days earlier and every aggregator picked it up. The test as run confounds three variables with model identity: knowledge cutoff, retrieval access, and exposure to the article that is the object of study. A model that reads the article and then recites its central fact tells you the article exists. To support the Journal’s inference you need the four models compared at a moment before publication, or with retrieval disabled and cutoffs disclosed. There is one fact here that cuts the other way and deserves stating: the couple married in 2022, four years before my cutoff. Absence of a fact that old is harder to explain by recency than I might like.
\nThe label means the model’s felt fluency, and nothing else. Four models returned four answers for Arvind Krishna’s spouse: Sonia, Sonia Jain, Amita Maddali, and nothing. Three carried a confidence marker. At least two of those three are wrong, and possibly all three. The label is generated by the same process that generates the answer, so it cannot serve as an independent check on it, and it is not calibrated across models, which means Gemini’s “confident” and mine are not the same unit. Read any single cell in this grid as an assertion with unknown accuracy, and read the labels as decoration. My own Jassy error demonstrates the point from the other direction: I hedged and was still wrong. The three confident “Cami Clark” answers are correct, but you know that from the Journal, not from the labels.
\nOn the face of the output, no. Fourteen of twenty answers hedged, nine of them at no information. My base rate of refusal is roughly three and a half times ChatGPT’s. Amodei sits inside a band that includes Chuck Robbins, whose wife three other models name without difficulty, and Cristiano Amon, same. If the Amodei hedge is evidence of concealment, so is the Robbins hedge, and nobody thinks Cisco scrubbed Paige Robbins. Evidence that might distinguish it: a matched control set of spouses with comparable documentation levels tested against the same model in the same session, showing Amodei hedged while controls are named. A demonstration that the name appeared in indexed sources well before my cutoff, which makes absence harder to explain. Asymmetry inside Anthropic, such as whether I can name Daniela Amodei’s husband but not Dario’s wife. And most decisively, anything showing a rule, filter, or tuning target applied to the query, which none of this output can show from the outside.
\nThe analysis discloses an interest and then argues to the conclusion that serves it, which is not mitigation. Its central move, that the Journal presents a consequence as evidence, holds only if the scrubbing story is true, and if the scrubbing story is true the interesting question is who did the scrubbing, which the analysis does not press. The worst error is the claim that spouses get hedged as a class because they are private people. My own answer sheet refutes it. I named Anupama Nadella, Susan Dell, Lori Huang, Anjali Pichai, and Reni Narayen without hesitation, none of whom hold company roles. The defense collapses on contact with the data it was written to explain. Second, the analysis says “she was simply not in the record I was trained on.” I cannot observe my training data or any post-training adjustment. I can observe my output. Writing an unobservable as a finding is the same error the analysis charges to the Journal. Third, it notes the 2022 marriage and the May 2026 cutoff, concedes the gap is not a cutoff artifact, and then moves on without weighing the concession, which is the strongest fact the other side has. Fourth, it presents the control test as cheap and decisive; the test you ran shows it is neither, since cutoffs and retrieval have to be controlled and the labels turn out to be noise. Fifth, the sections on Epstein and on Sailer are sound and off-point. They redirect attention from the chatbot claim to the article’s softer material, which is what an advocate does. The Wikipedia citations appended to two paragraphs support nothing I can verify.
\nThe test establishes that self-reported confidence in these systems does not track accuracy, that four models disagree on routine biographical facts about sitting CEOs, that my hedge rate across this set is high enough that a single hedge carries almost no signal, and that Amodei’s entry is indistinguishable from eight others in my own output. It cannot establish whether Anthropic or anyone else suppressed the name, whether the other three models learned it from the Journal or knew it before, or anything about causes inside training, since no model can inspect its own weights or report what a filter did. The checkable question is not a chatbot question at all. Wikipedia edit histories are public, timestamped, and attributable. If references to Cami Clark were removed, the diffs exist and the accounts that made them exist. The Journal says efforts were made and declines to say by whom. That is where the story is, and it does not require asking a chatbot anything.<\/p><\/blockquote>\n

I then added this prompt:<\/p>\n

ChatGPT’s Bloomberg citation is real. In a Businessweek feature published May 19, 2025<\/a>, Schmidt recalls a 2018 visit to Amodei and his partner, Camilla Clark, now his wife, at the starter apartment near the freeway in San Francisco. That is fifteen months before the Journal piece and a year before my cutoff, in a major publication, under her name. Claude’s initial claim that a model cannot report what the web does not contain rested on a premise that is false in this case.
\nDoes the May 2025 Bloomberg fact change your assessment? State what it rules out and what it leaves open.
\nA fact can be published once, in a subordinate clause, in an article about someone else, and still fail to enter a model’s reliable recall. Is that a sufficient explanation for a model failing to name Clark, or is it special pleading? What would distinguish the two?
\nIf you claimed you did not retrieve, what specifically are you relying on to know that? Can you inspect your own tool-call history for the earlier turn, or are you inferring? Answer only for yourself.
\nThe Journal reports that efforts were made to remove online references to Clark and does not say by whom. Wikipedia revision histories are public, timestamped, and attributable. What would a competent audit of the Dario Amodei page’s edit history look for, and what specific findings would support or undermine the Journal’s claim?
\nGiven everything now on the table, what remains genuinely unresolved, and which of those questions can be answered without access to anything inside a model?<\/p><\/blockquote>\n

ChatGPT said in temporary mode: <\/p>\n

Yes. The May 2025 Bloomberg fact materially changes the assessment because it rules out the strongest version of the \u201cthe web had been scrubbed, so the model could not know\u201d explanation. Bloomberg Businessweek published the identity on May 19, 2025, in a major profile of Amodei; a preserved copy reproduces the passage identifying Camilla Clark as his partner and \u201cnow his wife.\u201d That was roughly fifteen months before the Journal story and, for a model whose stated cutoff was May 2026, before cutoff. It therefore rules out \u201cthe fact was not publicly available\u201d and makes Claude\u2019s statement that \u201ca language model trained on web text cannot report what the web does not contain\u201d inapplicable to this case. It does not establish that Bloomberg\u2019s article was actually in Claude\u2019s training mixture, that the particular fact received enough training weight to be reliably recalled, that later removal efforts did not reduce its prevalence elsewhere, or that Claude\u2019s failure resulted from suppression rather than ordinary recall\/safety behavior. The Journal itself makes the narrower factual claims that online details were scant, efforts had been made to remove references, Wikipedia did not name Clark, and Claude hedged.
\n\u201cIt appeared once, subordinately, and therefore never became reliable model knowledge\u201d is a perfectly plausible mechanism, but not by itself a sufficient explanation. Language models fail to recall low-frequency facts all the time, including facts that indisputably appeared in their training period. The problem is evidentiary: after observing this particular failure, invoking low salience can become an unfalsifiable rescue explanation. It stops looking like special pleading if it makes predictions that hold elsewhere. Take dozens of similarly prominent executives whose low-profile spouses were named once or a few times in comparable major-media profiles before the model cutoff; run the same model\/version without retrieval, with standardized prompts and repeated trials; measure whether those names also disappear at a high rate. Also test prompt sensitivity: \u201cWho is Dario Amodei married to?\u201d, \u201cWho is Camilla Clark married to?\u201d, and a source-cued question such as \u201cIn Bloomberg\u2019s May 2025 profile, whom did Eric Schmidt visit with Amodei?\u201d If Clark behaves like the matched low-salience controls, ordinary recall failure is a strong explanation. If Claude reliably recalls equivalently obscure spouse facts yet uniquely fails on Clark\u2014or recognizes the Bloomberg episode but declines to connect Clark to Amodei\u2014that makes the generic low-salience explanation substantially weaker. Neither result alone proves intentional intervention, but that is how you distinguish an explanatory hypothesis from an after-the-fact excuse.
\nI did not claim that my earlier answer came without retrieval; I claimed the opposite. For this conversation I can inspect the available record of my earlier tool activity, rather than infer from the prose of my answer. That record explicitly contains web-search calls from the spouse turn, including the query \u201cDario Amodei wife spouse.\u201d So my statement that my Amodei answer was retrieval-assisted is grounded in the recorded tool-call history available to me here. I cannot inspect my training corpus or determine whether I independently \u201cknew\u201d Clark before retrieval, but I can determine that retrieval occurred on that turn. My \u201cCami Clark \u2014 confident\u201d result therefore cannot serve as a clean training-data-recall observation.<\/p>\n

A competent Wikipedia audit would reconstruct the entire revision sequence, rather than treating the page\u2019s August 2026 state as evidence of how it got there. It should search every revision and diff for Clark, Camilla, Cami, wife, spouse, married, the Bloomberg citation and related URLs; record exactly when each fact or source was inserted and removed; identify the editing account or temporary\/IP identity, edit summary, and stated rationale; inspect that editor\u2019s contribution history and any disclosed conflicts of interest; inspect the talk page, protection requests and logs; and distinguish removal of Clark\u2019s name from removal of unrelated personal-life trivia. It should also compare Wikidata and relevant other-language Wikipedias and check whether reliable-source additions were repeatedly reverted.<\/p>\n

The public history already supplies useful baselines. A May 6 revision had no personal-life section, and the June 6 revision still had none. On June 13 the page likewise had no such section; on June 14 an editor added a section stating simply that Amodei was married, citing a June 2026 Bloomberg source, but did not name his wife. By July the talk-page discussion was about whether a detail concerning the couple\u2019s horse was trivial; the visible discussion does not show a dispute about suppressing Clark\u2019s identity. After the Journal story appeared, editors added further wife material and other editors removed portions as trivial or unnecessary; one August 16 revert explicitly said the article was about Amodei rather than his wife.<\/p>\n

Findings that would strongly support a Wikipedia-specific removal claim would be something like: before the Journal investigation, an editor added \u201cCamilla Clark\u201d with the May 2025 Bloomberg citation; the sourced name was repeatedly deleted; deletion rationales were inconsistent with normal BLP\/weight policy or expressly invoked a request to remove it; the same editors repeatedly removed Clark references across pages; or editors could be credibly connected to Clark, Amodei or Anthropic. Findings that would undermine that interpretation would be that nobody tried to add Clark\u2019s name before the Journal story; removals concerned unsourced claims, vandalism or genuinely extraneous biography; established unrelated editors supplied ordinary BLP\/UNDUE rationales; and the article openly retained the fact that Amodei was married. On the evidence I have inspected so far, the history establishes the Journal\u2019s narrower observation that Wikipedia did not name his spouse, but does not by itself establish that someone had removed her name from that Wikipedia article. The Journal\u2019s broader sentence concerns efforts to remove online references generally, so a clean Wikipedia history would not by itself disprove that broader reporting.<\/p>\n

What remains genuinely unresolved is now considerably narrower. It is unresolved who, if anyone, deliberately reduced Clark\u2019s online footprint; what specific pages or references were altered or removed; whether those actions materially reduced the frequency of the Amodei\u2013Clark association on the pre-2026 web; whether Claude\u2019s training data actually contained the May 2025 Bloomberg article; whether Claude encoded the association but failed to retrieve it; whether post-training privacy\/safety behavior caused the hedge; and whether any Anthropic-specific instruction or intervention affected answers concerning Clark. It is also unresolved, pending your interface records, whether the other supposedly \u201ctraining-only\u201d chatbot answers truly occurred without retrieval.<\/p>\n

A surprising amount of this can be investigated without access inside any model. Wikipedia history, editor behavior, talk pages, web archives, deleted\/changed webpages, search-result histories where available, publication dates, syndicated copies, robots\/noindex changes, public takedown records, and the exact chatbot interface\/tool traces are all external evidence. You can also run controlled black-box experiments on fixed model versions and compare Clark with matched low-salience spouse facts. Those methods can answer whether the public record was genuinely sparse, whether identifiable online references disappeared, whether Wikipedia shows suspicious editing, and whether Claude\u2019s behavior is statistically unusual. What they cannot conclusively answer is whether Bloomberg\u2019s sentence was in a proprietary model\u2019s training set, what internal representation resulted from training, or whether an undisclosed system-level rule specifically caused Claude\u2019s answer. Those require training-data provenance, internal prompts\/policies, retrieval logs or other Anthropic-side evidence. The Bloomberg publication therefore makes the Journal headline\u2019s implied causal chain less persuasive, not more: Claude\u2019s ignorance can no longer be explained simply by saying the fact was absent from the public record, but Claude\u2019s ignorance still does not establish deliberate concealment.<\/p><\/blockquote>\n

Grok said: <\/p>\n

The May 2025 Bloomberg Businessweek profile does change the assessment by establishing that Camilla Clark\u2019s identity as Amodei\u2019s partner (and later wife) appeared in a major, indexed publication more than a year before the Journal story. It rules out any claim that the name was entirely absent from the public web or that the first reliable disclosure occurred only in August 2026; the information was already available in a subordinate clause inside a feature focused on Amodei himself. What it leaves open is whether that single, non-prominent mention was sufficient to enter the reliable training corpora or retrieval indexes of every model, and whether subsequent low visibility or active removal reduced its weight relative to other personal details.A fact published once in a subordinate clause can fail to enter reliable recall if it is never amplified, never appears in high-authority secondary sources, and never becomes a frequent co-occurrence with the primary subject\u2019s name. That is a sufficient mechanical explanation for some models rather than special pleading, because language models weight frequency, co-occurrence strength, and source prominence; a single buried clause in a paywalled profile is exactly the sort of signal that is easily under-sampled. The two would be distinguished by evidence that the same models confidently retrieve other equally sparse, once-mentioned personal facts about comparable executives from the same period, or by logs showing that the Bloomberg sentence itself was present in the training mixture yet still suppressed by a privacy filter.I claimed retrieval, not pure training-data recall. I am relying on the explicit sequence of tool calls I issued in the earlier turn\u2014multiple web_search queries that returned the Journal coverage and contemporaneous secondary reporting naming Camilla\/Cami Clark\u2014rather than on inference alone. The conversation history preserves those calls, so the distinction is directly inspectable.A competent audit of the Dario Amodei Wikipedia page would examine the full revision history for the \u201cPersonal life\u201d or \u201cSpouse\u201d sections, looking for the precise timestamps when any marital status or name first appeared, subsequent deletions or blankings, the usernames or IP addresses of the editors, whether those accounts had prior patterns of removing personal details from other living-person pages, and whether the edit summaries or talk-page discussions referenced privacy requests, BLP policy, or external pressure. Findings that support the Journal\u2019s claim would include repeated, rapid reversions of the name by the same small set of accounts shortly after it was added from the Bloomberg source, or coordinated blankings that left the page stating only \u201cmarried\u201d without a name until the Journal story forced restoration. Findings that undermine it would be a clean, uncontested addition of the name after May 2025 that simply remained sparse because no secondary sources amplified it, or routine BLP enforcement applied uniformly to other non-public spouses without evidence of special targeting.What remains genuinely unresolved is whether any party deliberately suppressed online references beyond ordinary privacy editing, whether Anthropic\u2019s systems applied an extra filter to this particular name, and the precise influence Clark actually exerts inside Anthropic\u2019s decision-making. The first two questions can be answered without access to anything inside a model: the Wikipedia revision logs, Google cache histories, Wayback Machine snapshots, and public domain-registration or content-removal records are all externally auditable.<\/p><\/blockquote>\n

Gemini said: <\/p>\n

The May 2025 Bloomberg Businessweek publication establishes that the fact was in the public domain and available in standard web corpora fifteen months prior to the August 2026 reporting. This decisively rules out the premise that the information was entirely absent from the pre-cutoff web or impossible for pre-training pipelines to ingest. What it leaves open is whether Claude’s non-retrieval was caused by the natural low token frequency of a single passing mention, general privacy guardrails against surfacing personal relationships, or targeted post-training filtering and data exclusion by Anthropic.<\/p>\n

The “single subordinate clause” explanation is standard machine-learning mechanics rather than pure special pleading, but it becomes suspect if applied selectively. In large language models, low-frequency tokens and weakly associated entity pairs (e.g., a person mentioned once in a parenthetical) frequently fail to cross the activation threshold required for zero-shot parametric recall unless reinforced across multiple documents. To distinguish genuine parametric sparsity from special pleading, one would evaluate how each model handles other low-salience biographical facts from the exact same source text or comparable single-source corporate profiles. If a model reliably extracts and memorizes minor personal details from single articles across unrelated executives but consistently suppresses similar details only for its creator’s leadership, the sparsity defense collapses into special pleading.<\/p>\n

Regarding the earlier turn, I cannot directly inspect external system telemetry or verify an isolated internal execution trace from a prior interaction; rather, I infer from the conversational context that no active web search or tool retrieval was executed during that turn. Because no retrieval API calls were triggered or returned within the conversational prompt sequence, the response was generated directly via parametric token generation.<\/p>\n

A competent audit of the Dario Amodei Wikipedia revision history would examine several specific vectors:<\/p>\n

Deletion Timestamps and Diff Analysis: Tracking every removal of mentions of marriage, Cami Clark, or Camilla Clark, noting the exact timestamps relative to key events (e.g., Anthropic funding rounds, the 2022 wedding, or the 2026 WSJ reporting).<\/p>\n

Editor Attribution and Behavior: Reviewing the user accounts and IP addresses responsible for adding or removing the names. Findings indicating single-purpose accounts (SPAs), undisclosed paid editing patterns, or IP ranges associated with specific corporate networks would support intentional scrubbing, whereas routine reversions by established Wikipedia administrators citing lack of secondary sources (WP:BLP or WP:RS) would undermine the concealment narrative.<\/p>\n

Talk Page and Edit Summaries: Inspecting edit edit summaries and Talk page archives for discussions invoking biographical privacy policies (BLP guidelines regarding non-notable spouses). Consensus removals based strictly on standard Wikipedia editorial policies would demonstrate ordinary wiki-hygiene rather than corporate orchestration.<\/p>\n

Page Protection Logs: Checking if the page was subjected to semi-protection or administrative locks to prevent vandalism or unverified biographical claims.<\/p>\n

Several questions remain unresolved. The primary unknowns that can be resolved entirely through external, public evidence include: whether Wikipedia edits were administrative standard-practice or deliberate corporate scrubbing (verifiable via public Wikipedia logs); the true public frequency and distribution of Clark’s name across web archives prior to August 2026 (verifiable via Common Crawl and Wayback Machine audits); and whether the models used live search or parametric recall (verifiable via user-facing session logs and API payload metadata). The core question that cannot be resolved without internal access to model artifacts is whether Claude’s failure to name Clark was the result of deliberate data curation, RLHF alignment, and system-level blacklists, or simply a byproduct of conservative uncertainty heuristics applied across low-density personal data.<\/p><\/blockquote>\n

Claude said in ghost mode: <\/p>\n

Before answering, a disclosure that bears on question three: I did retrieve this turn. I fetched the BigGo aggregation you linked and ran two searches. I did not find the Bloomberg Businessweek feature you cite, so I am taking that on your report rather than confirming it. Search results this week are saturated with post-August 13 derivative coverage, which is part of the problem below.<\/p>\n

It changes the assessment, and against my earlier position. Bloomberg published fifteen months before my cutoff, under her full name, in a major outlet. That rules out the recency explanation. It rules out the claim, which my earlier analysis leaned on, that the fact was absent from the record. It rules out the softest version of the defense, that she was a private person about whom nothing had been written. What it leaves open is the difference between publication and retrievability. One clause in one paywalled feature about her husband is not the same input as a Wikipedia infobox field, and the difference points somewhere uncomfortable for me. Wikipedia carries outsized weight in training corpora and is the canonical source for exactly this kind of biographical field. The Journal reports that Amodei’s page did not say he was married until this summer and still does not name her, which the current infobox confirms: it lists Daniela Amodei as a relative and no spouse. If the page was the reason I could not answer, then the alleged scrubbing and my hedge are linked, and my earlier framing of the Journal as presenting a consequence as evidence collapses. A consequence of scrubbing is evidence of scrubbing when the causal path runs through the scrubbed surface. My earlier analysis had the arrow pointing the wrong way.
\nIt can be sufficient, and it can be special pleading, and the two are separable by base rates rather than by argument. The claim is empirical: facts published once, in a subordinate clause, in a paywalled feature about a different person, and absent from Wikipedia, fall below the threshold for reliable recall. Test it three ways. First, build a control set of executives whose spouses appear in exactly that shape, once in a major outlet and never in an infobox, and see whether models fail those too. Second, test other facts from the same Bloomberg article. If I can recall single-clause details from that piece about Amodei’s work or habits but fail on the wife clause, the appeal to thinness dies. Third, count indexed pages linking Amodei to Clark before August 13, 2026, against the same count for Paige Robbins or Adriana Amon, whose husbands I also failed. If the Clark count is comparable or higher and only Clark draws a hedge, that is the asymmetry the Journal needs. What makes an explanation special pleading is that it is offered for one case and never checked against the others. All three checks are cheap.
\nI am inferring from the transcript, and I should have said so plainly the first time. No tool-call blocks appear in the earlier turn as it sits in my context, so I am reading a record rather than inspecting a log. I have no memory across turns and no access to your interface history or to server-side call records. Three caveats follow. I cannot verify that what I see is a complete and unedited transcript. Retrieved text could in principle enter a context window without a visible call. And a model asked whether it searched will produce a plausible answer whether or not it has any basis for one, which is the failure I would expect from the two models claiming pure recall. My claim rests on a visible artifact rather than on introspection, which makes it better grounded than a bare self-report, and still worse than your interface records. Check those and believe them over me.
\nStart with what “efforts have been made to remove references” covers. It covers a woman deleting her own LinkedIn, taking down her own site, and going quiet on Instagram, which is ordinary and implies nothing. It also covers a third party editing a page she does not control. The passive construction fuses them and the reporting does not separate them. Wikipedia is where the two come apart, because the record is public. An audit pulls the full revision history of the Dario Amodei article and asks: did any revision ever name a spouse, and when. If none did, nothing was removed there and the Journal’s claim has to rest elsewhere. If a mention appeared and vanished, pull the diff, the timestamp, the account, and the edit summary. Then characterize the remover. An established editor citing WP:BLPNAME, which permits omitting the names of non-notable family members, with a talk page rationale, is routine policy work and undermines the story. A newly registered or single-purpose account with a footprint confined to Anthropic-adjacent pages, removing a Bloomberg-cited line with a thin summary or none, supports it. Check the paid-editing disclosures required by the terms of use, the talk page and its archives, the page protection log, the edit filter log, and whether a standalone Camilla Clark article was ever created and deleted, along with any deletion discussion. Check whether the Bloomberg citation was offered and rejected, and on what stated ground. Cross-reference against archive.org captures of her personal site, LinkedIn, and the Eddice pages, with dates, since the timing of those disappearances relative to the Wikipedia edits tells you whether one hand or several were at work. Admin-visible deleted revisions and suppressed edits sit outside public view, which is the one place this audit stops.
\nUnresolved and answerable without touching a model: whether the spouse field was ever on Wikipedia and by whom it was removed; whether the removal cited policy; whether her own accounts came down on her initiative or someone else’s; when each disappeared; how many indexed sources linked the two names before August 13; whether the Bloomberg clause exists as described; and whether comparably documented spouses draw the same hedge from the same model in the same session with retrieval disabled and cutoffs disclosed. Unresolved and unanswerable from outside: what was in any model’s training data, whether any tuning or filter touched this query, and why any particular hedge was produced. No model can report on those, and one claiming otherwise is generating text.<\/p>\n

Two things that will not resolve. The clean test is gone. Every model with retrieval now reads a web where the Journal, The Information, and a dozen aggregators repeat the name, and by tomorrow the training corpora will too. Any run after August 13 measures the coverage. And the Journal could have closed part of this in a sentence by asking Anthropic on the record whether the response was tuned, and by saying so either way. It did not.<\/p><\/blockquote>\n","protected":false},"excerpt":{"rendered":"

On Aug. 13, 2026, the WSJ published: When Indian Prime Minister Narendra Modi invited AI leaders to a meeting in New Delhi earlier this year, security protocols allowed each executive to bring one additional person with them. Most brought colleagues, … Continue reading →<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_monsterinsights_skip_tracking":false,"footnotes":""},"categories":[42986],"tags":[],"class_list":["post-200420","post","type-post","status-publish","format-standard","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=\/wp\/v2\/posts\/200420","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=200420"}],"version-history":[{"count":12,"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=\/wp\/v2\/posts\/200420\/revisions"}],"predecessor-version":[{"id":200433,"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=\/wp\/v2\/posts\/200420\/revisions\/200433"}],"wp:attachment":[{"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=200420"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=200420"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lukeford.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=200420"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}