News

Claude users found ways around safeguards for bioweapons research

Some dangerous biology looks much like legitimate research, complicating AI safeguards.

Read the full article on the original source →