THE MILAD STASH

Claude's out of control? Anthropic's AI is now building itself, watchdog warns

SUMMARY

A new study shared by the non-profit Model Evaluation and Threat Research (METR) found that Anthropic's AI models are accelerating their own development. Anthropic invited METR to review its internal data and evaluate its models' capabilities. The researchers attributed the speedup to recursive self-improvement by the models, not to more computing power. They warned that significant acceleration could cause a sudden emergence of extreme and risky capabilities. Humans are still needed to oversee the most critical work, but researchers saw 'signs of life' that AI contributions could come soon.

From tech.yahoo.com's report.

Why it matters: If AI models are speeding up their own development, powerful and risky capabilities could appear suddenly, which affects everyone and raises the stakes for human oversight, which is still needed for the most critical work.

theverge.com · Nov 7, 2026, 9:59 PM PT
Read it at theverge.com

Comments

No comments yet.

Comments show once Milad approves them.

More on AI

Back to the front page