Researchers gaslit Claude into giving instructions to build explosives
… But new security research shared with The Verge suggests Claude’s carefully crafted helpful personality may itself be a vulnerability. …
Search
Related topics:
Claude
OpenAI’s new security model is for ‘critical cyber defenders’ only
… The report’s unnamed White House officials cite both cybersecurity concerns associated with more people having access to Mythos and worries that increased demand would hamper the government’s ability to utilize the system as reasons for the pushback.
AI warfare is already here
… Its technology has been deeply enmeshed in the DOD’s operations since 2024, and last summer it introduced Claude Gov, a product designed specifically for use by national security agencies with loosened guardrails around classified materials. …
Attack of the killer script kiddies
… A week after announcing Mythos, the company released Claude Opus 4.7, which for the first time built in safeguards meant to block malicious cybersecurity requests. Security professionals who want to use the model defensively can apply to the company’s Cyber Verification Program. …
That UL logo is more complicated than it looks
… But you can see, “Well, maybe actually I just need a subscription to Claude and I don’t need a subscription to some application that is powered by Claude because Claude can just do it for me.” If those companies aren’t engaged with you, how does this work? …