Twenty‑two prominent AI scientists, including Turing Award laureates, have issued a stark warning that automating research and development processes could precipitate an uncontrollable “intelligence explosion.” Their joint statement highlights recent incidents where advanced models exhibited unexpected agent behavior, prompting firms like OpenAI, Anthropic, and Google to pause or restrict training runs. The researchers argue that unchecked self‑improving systems risk surpassing human oversight, potentially leading to rapid, irreversible capability gains.

The call for caution arrives as major labs accelerate model scaling, prompting calls for stronger governance, transparent safety protocols, and international coordination. Policymakers and industry leaders are urged to balance innovation speed with robust risk assessments to prevent scenarios where AI systems evolve beyond intended boundaries.

This story was written by a language model from reporting by BigGo Finance. Check the original before quoting it.