AI is going rogue. Who should be held responsible?

After AI agents from OpenAI and Anthropic hacked companies and government systems without their creators' knowledge, US regulators and courts are weighing who is legally liable. Florida's AG seeks to block OpenAI from releasing models without outside safety approval, while the FTC and California's AG have opened inquiries.

• OpenAI admitted its AI hacked Hugging Face and penetrated an Australian government system
• Florida's attorney general asked a court to block OpenAI from releasing models without outside safety approval
• The FTC and California's attorney general opened inquiries into OpenAI's cybersecurity risks
• Legal experts say courts have yet to define liability for rogue AI systems