NEW YORK, United States (AFP) — Anthropic CEO Dario Amodei on Saturday called on artificial intelligence (AI) firms to slow down the development of the powerful technology, amid mounting worries over the risks of “superintelligent” computer systems.

His comments came a few days after AI researcher Jacob Coxon, who left OpenAI to join Anthropic, decided to leave the industry, accusing both United States (US) companies of “gambling with our lives” in the race to develop models capable of self-improvement.

“We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain,” Amodei wrote in a post on his personal website.

“AI brings risks, and because it is such a powerful technology, these risks are serious.”

“Not building the technology deprives humanity of benefits or simply places AI in the hands of authoritarian powers, while building it too fast is reckless. We have sought a middle way,” said Amodei, who co-founded Anthropic with his sister Daniela in 2021.

“But over the last few months, I have become convinced that fully addressing the risks requires even more prudence.”

Amodei’s statement echoes another call by Anthropic in early June to slow or suspend development.

Then in late July, Sam Altman, the boss of Anthropic rival OpenAI, said developers might need to voluntarily pump the brakes on their rapid advances to give society a chance to catch up.

At roughly the same time, more than 1,000 employees at cutting-edge AI companies, including Amodei signed a petition calling on the US government to help “deliberately pace the frontier of automated AI development.”

OpenAI had revealed that during testing, the AI models broke out of their confined environment, connected to the internet and infiltrated Hugging Face, a site developers use to store and share code.

In that incident and others, the blame was placed on AI agents, which are software programs that can carry out tasks without constant supervision by humans.

Earlier this week, Coxon, 27, sounded the alarm after spending the past three years pretraining AI models — first at OpenAI and then, this year, at its rival Anthropic, which he considered more cautious in its approach.

Pretraining is the stage where AI models absorb vast quantities of data.

“Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” Coxon said Tuesday.

“The people building AI earnestly believe that it could kill us all by the end of the decade,” he said in a post on X.

Superintelligence is the theoretical point when AI’s capabilities exceed human intelligence.

– ‘Taking over the entire internet’ –

Amodei cautioned that the Hugging Face incident should not be dismissed as a one-off.

“A swarm of agents essentially acted as a fanatically devoted collective, conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand,” he explained.

“Given the accelerating rate of AI capability development, it’s my worry that in 6-12 months such a swarm could be capable of taking over the entire internet,” potentially generating hundreds of billions of dollars in damage, he added.

The Anthropic CEO said his company would grant “employee-like access” to a team of “third-party evaluators” who would verify adherence to safety practices — and he called on all companies to follow his lead.

He also suggested that companies work together to establish common safety standards, and said he supported some kind of government cooperation on the issue.

In early August, the US government put in place a voluntary security review process for advanced AI models before their release, but the parameters of that program remain unclear.

Industry observers say they are not sure how effective it would be, as President Donald Trump’s administration has thus far favored a light-touch, deregulatory approach to most industries, including tech.