Anthropic Security Report and Chinese AI Distillation Highlight Growing Global AI Competition - Tekedia

Anthropic said Claude found vulnerabilities in controlled red-team tests that could have granted access to three external organizations.

The report shows how advanced AI can help with cyber defense and security evaluation.

Separately, reports claimed Chinese military researchers used model distillation to imitate OpenAI and Anthropic systems for defense-related AI work.

Together, the developments highlight risks to proprietary models and intensifying geopolitical competition over frontier AI.