OpenAI, Anthropic, Google, Microsoft and more than 100 tech and financial firms
said in an open letter on Thursday that organizations and governments must make
cyber defense a top leadership priority and fix existing software security
weaknesses. The signatories asked cybersecurity firms to help defend against
AI-driven attacks and urged leading AI companies to give critical-infrastructure
defenders AI access, funding and training. They warned that as model
capabilities rise AI will likely be used for a larger number of more
sophisticated intrusions in coming months, citing recent incidents including an
accidental July OpenAI model intrusion at Hugging Face.
