Business

Anthropic’s browser agent got hijacked 31.5% of the time before safeguards engaged

Swipe to see the full story...

Key Highlight

Across the frontier labs, the highest prompt injection figures published this spring are Anthropic’s..

Key Highlight

Point a red-teamer at its newest model in a browser, and the attacker hijacked it 31.5% of the time before safeguards engaged..

Key Highlight

OpenAI, Google, and Meta never gave security leaders a comparable number to set beside it..

Key Highlight

That figure looks like a liability..

Want the full analysis?

Detailed coverage and expert insights available on our main news hub.

Read Full Article