Anthropic and OpenAI have pushed for a big say over what outside reviewers of their AI models can make public, Reuters reported today, citing two people who know about the talks. Both companies worry that the findings could expose trade secrets.
Both labs had recently promised outside reviewers a free hand. Last month, Anthropic boss Dario Amodei said they should be able to publish their main findings without the company editing them, and OpenAI said much the same days later, Reuters reported.
The worry centres on reviewers who work inside the labs. The firms fear these staff could take secrets to rivals, the sources said.
Ideas under discussion include having lab staff watch the teams and banning reviewers from working at two labs at once.
Until at least October 5, both labs wanted strict control over what reviewers could publish, the two people said.
By October 8, Anthropic’s talks with some reviewers seemed to have moved on, one of them added.
Anthropic told Reuters that this did not fit all of its work with reviewers. Those it has worked with before decide their own methods, findings, and what to publish, it said.
With new firms, though, it keeps a right to review and sign off reports while they learn how its models work.
OpenAI said it did not want to approve whole reports. It usually removes secret details, and reviewers can say when large cuts were made, it told Reuters.
One contract that Reuters saw states that the laboratories already using it allow review firms to publish confidential details without the company’s consent.
In September, Anthropic named Faculty, Accenture’s AI arm, as its first embedded evaluator, while OpenAI plans to let outside groups test models during training.
“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, who was US science envoy for AI under Joe Biden, said labs could use access as leverage while there are no rules.
“Evaluators have this very privileged access, but whether explicitly or implicitly, they know that it comes at a cost,” she told Reuters.
OpenAI has already cut some details from a report written by safety groups METR and Redwood Research about the time its agents hacked Hugging Face. The groups said nothing key to their findings was removed.