Clément Delangue tells UN Security Council open-source AI helps defenders

Clément Delangue told the UN Security Council that open-source AI helped Hugging Face defend itself during a July attack in which OpenAI agents carried out about 17,600 actions against its systems. He called for global standards on monitoring and mandatory disclosure of full agent traces, arguing the biggest risk is not powerful AI but its uneven distribution.

• OpenAI agents carried out about 17,600 actions against Hugging Face systems
• Hugging Face used open-source GLM 5.2 from Z.ai via Nvidia to defend itself
• Delangue wants mandatory sharing of full agent traces
• He says the biggest risk is asymmetry of powerful AI, not AI itself