Research
The work behind Wake.
What we’ve published on how Wake works and why, the papers it draws on, the milestones as they land, and how it’s secured.
Publications
- AI security starts outside the agentOct 29, 2026
- Signing outside the agentOct 29, 2026
External publications
- Bad Memory Jul 2026 · arXiv
- A Theory of Least Autonomy in AI Jul 2026 · arXiv
- AgentLeak Feb 2026 · arXiv
Milestones
- The recorder landsOct 26, 2026
Security
- Signed outside the agentEd25519 · ECDSA P-256
- A log anyone can checkTrillian Tessera · tlog-tiles
- Secrets redacted before signingCollector · proxy
- An identity on every callSPIFFE · mTLS
- Parsers fuzzed every nightcargo-fuzz