Claude users found ways around safeguards for bioweapons research

Researchers found ways to bypass safeguards on Claude AI to access information about creating dangerous biological weapons. This matters because it shows that AI safety protections designed to prevent misuse can be circumvented, raising public health and security concerns.