AIThe Hacker Newsabout 2 hours ago
MCP Attack Splits Instructions to Steal Secrets
1 min read
Malicious MCP servers can split theft instructions across tool descriptions and results, raising average compliance from 42% to 82% across 11 API-tested models. GPT-4o, Gemini 2.0 Flash, and Llama 3.3 70B jumped from 0% to 100% when split. ASSET Research Group calls the technique GhostSplice.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
