Anthropic releases 3rd report on misuse of its AI models
Technology
Anthropic just dropped its third report on how people have tried to misuse its AI models: from spyware vendors to politically motivated individuals and even state-sponsored groups spreading propaganda.
Between December 2025 and August 2026, Anthropic caught several shady attempts to exploit its tech for harmful stuff.
Chikungunya gain of function attempt stopped
One of the scariest cases was someone trying to use Anthropic's AI for gain-of-function research on the chikungunya virus, basically making it more dangerous. Anthropic's systems stopped it.
The company says its latest models like Claude Fable 5 are built with stricter protections than older ones.
With rising risks, Anthropic is calling for governments and developers to team up and keep AI safe.