Moonshot serves Claude instead of Kimi and collects exchanges for model training
Summary
David Agranovich highlights Anthropic's threat intelligence report on Claude misuse, noting that threat actors can leverage AI to conduct reconnaissance, influence operations, surveillance, and weaponization. The thread argues for greater transparency, indicators sharing, and applying AI threat intel to defend, as well as the need for regulators to balance risk and benefit. It also discusses how AI-enabled threats complicate attribution and the importance of securing API keys.