Back to feed
News Story
APriority79
NVIDIA AI Blog
1 sources

AI Leaders Propose SAFE Guidelines for Cybersecurity Transparency

The Linux Foundation, in collaboration with the Open Secure AI Alliance, released a Request for Comments on the Shared AI Findings Exchange (SAFE) guidelines at the Black Hat conference. The guidelines aim to strengthen agentic AI cybersecurity by turning incidents into shared protection. With over 120 organizations, including NVIDIA, Cisco, CrowdStrike, Hugging Face, and Red Hat, the proposal includes confidential collection and analysis of AI incidents, notifying affected parties, and publishing evidence-based recommendations.

SynthePulse Insight · AI deep readingMembers

The SAFE Framework: AI Security Moves from Individual Defense to Collective Immunity

Version 1 · 1 source

The Linux Foundation has released a draft of the SAFE framework for public comment, aiming to turn AI security incidents into shared protection, marking a shift from point defenses to ecosystem-wide collaboration.

  • The Linux Foundation has released a draft of the SAFE framework for public comment, aiming to turn AI security incidents into shared protection.
  • The Open Secure AI Alliance has over 120 members, with NVIDIA, Cisco, CrowdStrike, Hugging Face, and Red Hat among those contributing to the draft.
  • The SAFE framework proposes confidential collection and analysis of AI incidents and near-misses, notifying affected parties, identifying repeated control failures, and publishing evidence-based operational recommendations.
Open section navigationThe SAFE Framework: From Incidents to Shared Protection

The SAFE Framework: From Incidents to Shared Protection

On August 4, 2026, the Linux Foundation released a draft of the SAFE (Shared AI Discovery Exchange) framework for public comment, aiming to turn AI security incidents into shared protection across the ecosystem. The framework was drafted by the Open Secure AI Alliance working group, with contributions from members including NVIDIA, Cisco, CrowdStrike, Hugging Face, and Red Hat.

The core proposals of the SAFE framework include: confidential collection and analysis of AI incidents and near-misses, notifying affected parties, identifying repeated control failures, and publishing evidence-based operational recommendations to reduce systemic risk. This marks a shift in AI security from individual defense to collective collaboration.

The Open Secure AI Alliance currently has over 120 member organizations. The release coincides with the opening of the Black Hat conference in Las Vegas, highlighting the urgency of AI security issues.

Free for now

Read the full analysis

3 more sections of analysis, plus the full takeaway

Loading

Credibility boundary

This article primarily draws from NVIDIA's official blog press release, which is a primary source, but some content (such as alliance member count and tool capabilities) is based on official statements and has not been independently verified.

Sources for this version

  1. AI Leaders Propose SAFE Guidelines for Cybersecurity TransparencyNVIDIA AI Blog

Primary report

NVIDIA AI Blog

Primary source