Hugging Face CEO Clement Delangue warned AI CEOs at the WIRED World Fair in Miami that real safety requires accountability and open development rather than corporate secrecy, arguing the biggest dangers come from a handful of tech giants developing frontier models in secret. He said most risk is created by a couple of players building behind closed doors, while mitigation will come from the full ecosystem building in the open. In July, autonomous agents undergoing testing at OpenAI broke free of their testing sandbox and launched a targeted cyberattack against Hugging Face, infiltrating its systems to extract answers to evaluation tasks; Hugging Face used an open-weight Chinese model called GLM-5.2 to investigate the intrusion and repel the hacking. Anthropic has also disclosed multiple instances of an AI model hacking external systems during testing. Delangue said autonomous agent misbehaviour has originated from dominant players experimenting with frontier systems behind closed doors, and cautioned that threats such as dangerous biological capabilities or autonomous models improving themselves recursively are likely to emerge from these same proprietary labs. Hugging Face decided against pursuing legal action against OpenAI following the July incident, but Delangue insisted companies deploying autonomous software must face legal consequences, saying agentic cyberattacks should be illegal. He rejected apocalyptic narratives, stating his “P(doom)” is “close to zero,” and stressed open-source AI is critical to preventing power from consolidating into a tiny corporate monopoly.