AIThe Hacker Newsabout 6 hours ago
MemGhost Attack Plants False AI Memories
1 min read
A single email can trick an AI agent into saving a false memory that persists across sessions. Researchers built MemGhost, an automated tool that writes the malicious emails. The attack succeeded in 87.5% of background-mode tests against OpenClaw on GPT-5.4 and 71.4% against a Claude Code SDK agent on Sonnet 4.6.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
