Brief · Source report Claude users found ways around safeguards for bioweapons research Ars Technica AI 2026-09-11 Back to home Read original Some dangerous biology looks much like legitimate research, complicating AI safeguards. Continue on the homepage →