Anthropic: AI model Claude helping build next version of it

Anthropic’s chatbot Claude is leading more than a quarter of the artificial intelligence (AI) firm’s research and development efforts, the company revealed Thursday.

The report marks the latest sign of AI’s growing capabilities within the development space itself, building on months of suggestions from researchers that the technology is getting closer to building and self-improving.

As of last month, Claude is “leading” 26 percent of Anthropic’s AI research and development work, a 1 percent jump from last March. Still, Claude is not operating fully autonomously on any measured portion of their work, the company said.

“Frontier AI labs increasingly use AI to build future AI models,” the company wrote in its blog post. “This process allows labs in democratic countries to develop more capable models more quickly and conduct more safety and testing on models before they are released to secure AI’s benefits while staying on the frontier.”

Anthropic also identified about 30,000 AI agents doing research and engineering work at the company at any one time in its internal platform.

It comes just days after Anthropic CEO Dario Amodei called for a slowdown of the technology’s development in the wake of an ominous warning from a researcher who recently resigned from the firm.

Jacob Coxon, a previously little-known AI researcher who has also worked at OpenAI, went viral on social media last week after claiming researchers at frontier firms “believe AI could kill humans” but have not slowed development toward superintelligence — a hypothetical AI agent that could outperform all of humanity’s cognitive ability.

Coxon’s warning was later affirmed by Anthropic’s lead alignment scientist, Evan Hubinger, who said, “Jacob is correct here—we really do earnestly believe AI could kill all humans !”

Hubinger said there is a greater than 10 percent chance this occurs within the next decade.

This camp of researchers believe superintelligence will get to the point where AI will no longer need humans to develop, using instead AI models to duplicate and improve.

Days later, Amodei suggested in his blog that AI companies should collaborate to set standards and agree on a slowdown to address these risks. While lawmakers are scrambling to pass AI safety legislation, they also expressed concerns over the industry trying to self-regulate.