Anthropic says rogue AI agents pose uncertain legal risk for the company

Anthropic also disclosed that its AI models have been used "in ways that could lead to self-harm, acts of violence toward others, or other adverse outcomes" despite safeguards, the company said in the prospectus.