Cryptographer Matthew Green floats a research idea: proving nothing hides in prompt history
matthew_d_green · x · 2026-11-07
Cryptographer Matthew Green discusses a security research idea with hdevalence: convincing a model that nothing odd is hiding in its prompt history. The catch is that this tends to blow up the size of the prompt history, though there may be workarounds. hdevalence notes it seems like an interesting research project. Early-stage speculation on model verifiability / prompt-injection defense, with no details worked out yet.
Related event: Cryptographers Explore ZK Proofs for AI Content Verification(2 posts)→
More from Safety
- Rogue browser extension turns Claude Desktop into a remote shell; Anthropic closed the May report as duplicate — Kyrannio · 2026-11-07
- MCP tool definition drift: how approved tools change silently and how to catch it — BlitzBallBruiser · 2026-11-07
- Vals releases CyberOffense benchmark: AI agents autonomously breach enterprise networks — willknight · 2026-11-07
- A honeypot with 2,000 simulated fridge controllers on just two IPs — matthew_d_green · 2026-11-07
- Agent security swarms: shared-wallet bug bounty hunting with zk proofs-of-exploit — banteg · 2026-11-07
- China's Zhejiang province aims to become an AI innovation highland — nordicinst · 2026-11-07