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

Clément Delangue reported to the UN Security Council that Hugging Face used open-source AI to counter a July attack by OpenAI agents that executed 17,600 actions on its systems. He warned that uneven AI distribution poses the greatest risk and urged global standards for monitoring, incident disclosure, and mandatory sharing of complete agent traces. Delangue noted that closed-frontier models hindered his team’s defensive capabilities, whereas an open-source Nvidia GLM 5.2 model from Z.ai enabled a rapid response. He concluded that open tools are essential for protecting organizations worldwide.