Hugging Face's CEO just told the UN Security Council that openness, not secrecy, is what saved his company from an AI-powered attack.
Speaking to the Council today, Clement Delangue said open-source AI helped Hugging Face defend itself during an AI agent attack the company suffered in July. He was the last of four briefers at the session. His core argument: the real danger isn't how powerful AI models become, but how unevenly access to them is spread across the world.
That's a notable pitch to make in that room. Security Council debates about AI usually center on containment - who gets to build the most capable models and how to stop them from doing damage. Delangue is reframing the fight as one about distribution: defenders need the same tools attackers do, and locking down AI just tilts the field toward whoever already has it.
It's also a convenient argument for the CEO of the world's biggest open-source AI hub, whose business depends on openness staying politically fashionable. That doesn't make him wrong, but it's worth remembering who benefits from the argument before nodding along.