UN’s AI experts warn against ‘apocalyptic’ rhetoric on risks

Several experts on a UN artificial intelligence committee cautioned against sensational claims about the dangers of AI during a recent event in New York. Former AI researchers from Anthropic and Google DeepMind expressed concern that some professionals were using apocalyptic language to describe potential risks associated with AI development.

Joelle Barral, a Google DeepMind executive and member of the International Scientific Panel on AI, argued that fear-mongering is unhelpful and that scientists should focus on understanding both what is known and unknown about AI risks. She highlighted the importance of avoiding unsubstantiated claims that AI could lead to the eradication of humanity by the end of the decade.

Nobel Peace Prize laureate Maria Ressa echoed this sentiment, emphasizing that the discourse around AI risks needs to move beyond simplistic binary assessments. The International Scientific Panel on AI, established in 2025, aims to enhance scientific knowledge about AI and guide UN discussions on the matter. Panel co-chair Yoshua Bengio stressed the need for nuanced and sophisticated discourse, distinguishing between scientifically verified knowledge and speculative extrapolations about AI risks.

Bengio warned against the pitfalls of overly dramatic rhetoric, advocating for a global, rational conversation that acknowledges both the uncertainties and severity of potential AI risks.