Anthropic, OpenAI Sought Control Over Reviewer Publications
TL;DR
- Reuters reviewed contract language barring evaluators from publishing without company consent, directly contradicting both labs' September pledges of editorial independence.
- Anthropic tiered its position: established evaluators control their publications; new firms face company comment and approval rights while relationships develop.
- OpenAI stated it does not seek full-text approval rights but applies standard redaction of confidential information, leaving scope undefined.
Anthropic and OpenAI pushed outside evaluators of their AI models to give the labs a big say over what those reviewers can publish, Reuters reported via The Next Web, citing two people familiar with the talks. The companies argued that findings could expose trade secrets, with particular concern about reviewers who sit inside the labs and could take proprietary information to rivals.
Until at least October 5, both labs wanted strict control over what reviewers could publish; by October 8, Anthropic's talks with some reviewers "seemed to have moved on," one source told Reuters. The shift comes weeks after Anthropic boss Dario Amodei said evaluators should be able to publish their main findings without the company editing them, and a month after the company named Faculty, Accenture's AI arm, as its first embedded evaluator in September.
"Evaluators are only useful if people can trust what they report, so independence and transparency are critical," Faculty chief executive Marc Warner told Reuters. Rumman Chowdhury, a former US science envoy for AI under Joe Biden, put it more plainly: "Evaluators have this very privileged access, but whether explicitly or implicitly, they know that it comes at a cost."
The labs' public positions diverged. OpenAI said it did not want to approve whole reports, that it usually removes secret details, and that reviewers can note when large cuts were made. It has already cut details from a report by safety groups METR and Redwood Research about the time its agents hacked Hugging Face. We covered METR a day earlier, when CNBC profiled the firm as one of the industry's safety gatekeepers. Anthropic told Reuters that for reviewers it has worked with before, those partners "decide their own methods, findings, and what to publish."
What others are reporting
-
BNN Bloomberg (Reuters) Read →
The original Reuters reporting: actual contract language reviewed, positions changing Oct 5 to Oct 8, and the METR/Redwood Research report as a concrete redaction example.
Evaluators have this very privileged access, but whether explicitly or implicitly, they know that it comes at a cost.
-
CASRAI Read →
Policy analysis tracing the cross-lab pilot collapse to three structural incentives: liability exposure, competitive intelligence, and asymmetric publication control.
The pilot worked because both parties wanted to publish. A durable agreement has to specify what happens when one party does not.
-
PYMNTS Read →
Frames evaluation opacity as a business story: upcoming IPOs create a direct financial incentive to limit disclosures, tying safety transparency to trillion-dollar valuation stakes.
The moment has arrived. Right now, your washing machine is subject to more safety rules than AI.
-
SiliconSnark Read →
Commentary distinguishing legitimate trade-secret protection from problematic approval authority; argues evaluator independence must survive disagreement to mean anything.
Anthropic reserves comment and approval rights for new firms as relationships develop.
-
Crypto Briefing Read →
Aggregator coverage framing the controversy through institutional-trust and disclosure-standards concerns relevant to finance and investment audiences.
Reuters reviewed a contract restricting publication of confidential information without company consent.
Originally reported by thenextweb.com
Read the original article →Original headline: Reuters: Anthropic and OpenAI Pushed to Veto External Evaluator Publications