Guardrails off for Anthropic: Firm tweaks AI safety policy amid heightened competition, lack of regulation—what changes?

The guardrails are off for Anthropic, a company founded by former OpenAI employees worried about the dangers of artificial intelligence.

Once focused on the proper development of AI technology with safety in mind, Anthropic is now weakening its foundational safety principle, with the company releasing a statement on its revised security policy.

“We’re releasing the third version of our Responsible Scaling Policy (RSP), the voluntary framework we use to mitigate catastrophic risks from AI systems,” Anthropic said on Tuesday, marking the change.

Anthropic’s new safety policy: What changes?

In a statement to Business Insider, the company also said that amid heightened competition and lack of government regulation, it would no longer abide by its commitment “to pause the scaling and/or delay the deployment of new models” when such advancements would have outpaced its own safety measures.

Anthropic’s previous safety policy required it to pause training more powerful models if their capabilities outpaced the company’s ability to control them and ensure their safety — a measure that’s been removed in the new policy.

Explaining the shift, Anthropic said that the current policy environment with regard to the technology had “shifted toward prioritizing AI competitiveness and economic growth, while safety-oriented discussions have yet to gain meaningful traction at the federal level.”

Further, the company’s chief science officer, Jared Kaplan, told Time Magazine that its responsible scaling policy had failed to keep pace with the AI race.

“We felt that it wouldn’t actually help anyone for us to stop training AI models. We didn’t really feel, with the rapid advance of AI, that it made sense for us to make unilateral commitments … if competitors are blazing ahead,” Kaplan was quoted as saying by the publication.

Anthropic also said that it was “convinced” that “effective government engagement on AI safety is both necessary and achievable”, but added that it was “proving to be a long-term project—not something that is happening organically as AI becomes more capable or crosses certain thresholds.”

To that end, Anthropic will continue to provide safety recommendations for the AI industry, but the company will separate its own plans from its suggestions for the industry.

Tiff with Pentagon

The change comes at a time when Anthropic has been embroiled in a dispute with the Pentagon, and a day after Defense Secretary Pete Hegseth gave the company’s CEO Dario Amodei a Friday deadline to rollback AI safeguards.

Failing to do so, Hegseth warned, would put Anthropic at risk of losing a $200 million defence contract and being put on a government blacklist, reported CNN.

That said, a source familiar with developments told news outlet that the change in Anthropic’s safety policy was not related to the Pentagon case.