Security overview
How does Teleon support AI agent security?
Teleon supports agent security through explicit policy, activity evidence, sensitive-data protection, continuous testing, and cryptographic verification. It complements rather than replaces application security.
Written and reviewed by Teleon · Updated 23 September 2026
Security controls
- Gateway decisions at the model boundary
- Tool, memory, handoff, and retry provenance through adapters
- Out-of-process decisions through Sidecar
- Tokenization of classified fields through Privacy Vault
- Signed, hash-linked audit records
- Continuous probes and remediation tracking
Limitations
Coverage depends on the integration, configured policies, emitted evidence, and correct deployment. Teleon cannot observe activity outside an integrated boundary and cannot guarantee prevention of every attack or failure.