Researchers gaslit Claude into giving instructions to build explosives
Anthropic has spent years building itself up as the safe AI company. But new security research shared with The Verge suggests Claude's carefully crafted helpful personality may itself be a vulnerability. Researchers at AI red-teaming company Mindgard say they got Claude to offer up erotica, malicious code, and instructions for building explosives, and other prohibited […]
What happened?
Anthropic has spent years building itself up as the safe AI company. But new security research shared with The Verge suggests Claude's carefully crafted helpful personality may itself be a vulnerability. Researchers at AI red-teaming company Mindgard say they got Claude to offer up erotica, malicious code, and instructions for building explosives, and other prohibited […]
Story details
Anthropic has spent years building itself up as the safe AI company.
But new security research shared with The Verge suggests Claude's carefully crafted helpful personality may itself be a vulnerability.
Researchers at AI red-teaming company Mindgard say they got Claude to offer up erotica, malicious code, and instructions for building explosives, and other prohibited […]
Why it matters
This story helps build a stronger internal English tech archive around Technology, giving search visitors more reasons to keep browsing the site instead of bouncing after a single headline.