From Trust Assumptions To Trust Evidence: Why PKI and Confidential Computing
summary
DigiCert demonstrates two concrete solutions: retrofitting TPM-based measured boot and model integrity checks onto deployed MRIs using device identity certificates when TPMs are absent, with graduated security, key escrow/release, ledger-backed signed attestations, and handling offline devices and CPU-constrained firmware; and issuing an AI agent passport—a short-lived JSON carrying SVIDs, attestations, policy and lineage (SPIFFE-inspired) to federate and revoke agent identity. A preview deployment is live with customers.
AI-written summary · about
chapters (13)
- 0:00Trust assumptions and evidence
- 1:04Regulated industries' competing interests
- 3:09Healthcare AI deployment stakeholders
- 5:15Confidential computing for distrust
- 7:21Greenfield vs brownfield protection
- 9:46Trust in existing medical devices
- 11:08Collecting and verifying trust evidence
- 12:30Enforcing trust with degradation modes
- 15:18Actor identity in AI environments
- 17:21AI agent passports for workload identity
- 20:12Gaps in AI trust frameworks
- 21:55Customer preview and future
- 22:37Integrated trust solution
related talks
56mIoT: Taking PKI Where No PKI Has Gone Before
35mChains of trust in Confidential Computing
29mRemote Attestation of IoT devices using a discrete TPM 2.0
29mA Simple Protocol for Remote Attestation of System Integrity
36mSecurity for AI and Multi-Party Collaboration with Confidential Computing (Oh, and...
37mIt's All About Trust: Why Confidential Computing Matters
40mDon’t Trust Opaque Clouds, Cryptographically Verify Instead!
49mKeylime: Bootstrapping and Maintaining Trust