UN AI panel experts push back against doomsday warnings

Members of a United Nations artificial intelligence committee have cautioned against extreme language used by some industry figures concerning AI dangers. Speaking in New York on September 21 on the sidelines of the UN General Assembly, experts urged a clearer separation between proven facts and extreme fears.

Expert Warnings

Joelle Barral, a Google DeepMind executive and member of the International Scientific Panel on AI, stated that investing in fear is unhelpful and that scientists must focus on understanding real knowledge gaps. Nobel Peace Prize laureate Maria Ressa added that public discussions often swing too far between total doom involving killer robots and dismissing risks entirely. Yoshua Bengio, a co-chair of the panel, emphasized the need to separate what researchers actually know from what is only guessed through plausible extension.

Industry Context

These warnings follow recent departures by tech workers. Former Anthropic and OpenAI researcher Jacob Coxon left the industry in early September, stating that developers believed the technology posed extreme fatal risks. Another research engineer who exited Google DeepMind in July similarly claimed it might already be too late to prevent human eradication. The International Scientific Panel on AI was established in 2025, with members chosen in February 2026, to improve scientific understanding and support UN discussions.