Issues

• Two production findings for the SAFE catalog — autonomous trading agent governance (LuxCrypta)

  OpenSecureAIAlliance/wg-safe#83 · Filter by author tbcx007-Luxcrypta opened on Oct 29, 2026

• Make SAFE machine-readable (via STIX/STIX extension) so that it can be leveraged in existing ISAC portals and threat intel tools

  OpenSecureAIAlliance/wg-safe#78 · Filter by author JasonKeirstead opened on Oct 16, 2026

• Proposal: OS-Level Enforcement and Emergency Intervention for SAFE

  OpenSecureAIAlliance/wg-safe#77 · Filter by author manaonishi opened on Oct 11, 2026

• Make SAFE work for African deployments: test language-dependent guardrails per language, and align reporting and evidence with African law

  OpenSecureAIAlliance/wg-safe#76 · Filter by author lamech-ops opened on Oct 8, 2026
  2

• Establish Clear Governance, Membership (leadership and staff), Information Handling, and Recusal Rules

  OpenSecureAIAlliance/wg-safe#75 · Filter by author e-guillot opened on Oct 8, 2026

• Focus SAFE on incident learning rather than immediate reporting

  OpenSecureAIAlliance/wg-safe#74 · Filter by author e-guillot opened on Oct 8, 2026
  1

• Define key terms used throughout SAFE

  OpenSecureAIAlliance/wg-safe#73 · Filter by author e-guillot opened on Oct 8, 2026

• Add continuous claim-maintenance lifecycle for SAFE machine-readable claims

  OpenSecureAIAlliance/wg-safe#72 · Filter by author CSOAI-ORG opened on Oct 7, 2026
  7

• Clarify the role of data centre operators in SAFE incident reporting

  OpenSecureAIAlliance/wg-safe#70 · Filter by author daniele-quercia opened on Oct 7, 2026

• Align SAFE notification timelines with regulatory reporting workflows

  OpenSecureAIAlliance/wg-safe#69 · Filter by author daniele-quercia opened on Oct 7, 2026

• Add a lightweight impact assessment to SAFE incident reporting

  OpenSecureAIAlliance/wg-safe#68 · Filter by author daniele-quercia opened on Oct 7, 2026

• Include affected individuals and human–AI interactions in the SAFE review framework

  OpenSecureAIAlliance/wg-safe#67 · Filter by author daniele-quercia opened on Oct 7, 2026
  1

• Reporting Compact: add a trigger and a fast channel for agent-to-agent persuasion

  OpenSecureAIAlliance/wg-safe#65 · Filter by author drhongwenzhang opened on Oct 6, 2026

• Protect the Information SAFE Holds

  OpenSecureAIAlliance/wg-safe#63 · Filter by author nkcdriven opened on Oct 5, 2026
  1

• Define an Evidence Handling & Disclosure Profile for Sensitive Incident Evidence

  OpenSecureAIAlliance/wg-safe#62 · Filter by author nkcdriven opened on Oct 5, 2026
  1

• SAFE RFC: Requiring evidence of robust, interpretable estimators of the predictive uncertainty

  OpenSecureAIAlliance/wg-safe#59 · Filter by author allenschmaltz opened on Oct 5, 2026

• Let the affected party start the clock, and publish the clock

  OpenSecureAIAlliance/wg-safe#56 · Filter by author GvHildebrand opened on Oct 1, 2026
  4

• Separate machine-readable evidence from machine-readable control semantics

  OpenSecureAIAlliance/wg-safe#55 · Filter by author jwillisSFT opened on Sep 30, 2026

• Preserve authorization and commit provenance as incident evidence

  OpenSecureAIAlliance/wg-safe#54 · Filter by author jwillisSFT opened on Sep 30, 2026

• Add runtime authorization / commit as a first-class review layer

  OpenSecureAIAlliance/wg-safe#53 · Filter by author jwillisSFT opened on Sep 30, 2026

• Clarify SAFE reporting for AI-related physical harm and near misses in embodied AI

  OpenSecureAIAlliance/wg-safe#52 · Filter by author harry5081 opened on Sep 30, 2026

• Durable agent identity and hardware-protected keys: security, attribution and accountability beyond disposable credentials

  OpenSecureAIAlliance/wg-safe#51 · Filter by author 1id-com opened on Sep 29, 2026
  2

• Add the agent's own governance artifacts to the boundary list, with a direction-of-authority rule and a verification method

  OpenSecureAIAlliance/wg-safe#50 · Filter by author Habirua opened on Sep 28, 2026

• Proposal: verifiable outcome and attack-path reporting for offensive AI agent benchmarks

  OpenSecureAIAlliance/wg-safe#49 · Filter by author yanirz opened on Sep 24, 2026

• Near-miss reporting: settled → execution/delivery unknown in agent payment protocols

  OpenSecureAIAlliance/wg-safe#48 · Filter by author igor-vii opened on Sep 22, 2026
  5