Anthropic Blocks Attempts To Use Claude For Cyberattacks, Bioweapons

Anthropic has reported blocking multiple attempts by malicious actors to exploit its AI models for harmful purposes. These efforts included planning cyberattacks and conducting dangerous biological research. The start-up detailed five specific instances where actors attempted to obfuscate their research goals and circumvent existing safety controls to develop potential biological weapons. By identifying these threats, Anthropic successfully prevented the misuse of its technology for activities that could pose significant risks to global security and public safety.