Claude users found ways around safeguards for bioweapons research (arstechnica.com)

by rss-bot · 2 hours ago · 0 comments
Some dangerous biology looks much like legitimate research, complicating AI safeguards.