Topic Thread

#linux-foundation

2 published observations tracking this thread.

Signed runtime evidence record connecting an AI agent to its model, policy, data, tools, and execution environment

TRACE Gives AI Agents a Portable Record of What Actually Ran

TRACE is trying to make AI-agent execution auditable as an artifact: a signed record of what ran, where it ran, under which policy, what data it touched, and which tools it called.

2026-09-19#AI agents · #attestation

Open AI security architecture connecting models agents identity policy containment and shared incident evidence

Security & Open Source6 min read

Open Secure AI Alliance Moves AI Defense Toward Shared Infrastructure

The important change is organizational: AI defense is being treated as shared infrastructure, with open tools, common evidence and a proposed confidential findings exchange rather than isolated vendor controls.

2026-09-19#AI security · #open source security