AI Leaders Propose SAFE Guidelines for Cybersecurity Transparency

At the Black Hat conference in Las Vegas, the Open Secure AI Alliance, now over 120 organizations strong, is developing new guidelines to strengthen agentic AI cybersecurity. The Linux Foundation shared a Request for Comments on Shared AI Findings Exchange (SAFE), a proposed set of guidelines designed to turn agentic cybersecurity incidents into shared protection. SAFE is being drafted by an alliance working group, with contributions from NVIDIA, Cisco, CrowdStrike, Hugging Face, and Red Hat. The guidelines include proposals to confidentially collect and analyze AI incidents and near misses, inform those impacted, identify recurring control failures, and publish evidence-based operating recommendations. NVIDIA contributed several open-source tools and models, including NOOA, OpenShell, Nemotron, Cosmos, Isaac GR00T, BioNeMo, Alpamayo, NeMo Guardrails, NeMo Anonymizer, and Garak. Other members made contributions across identity, harnesses, models, and availability layers, including Okta, Palo Alto Networks, Red Hat, Amazon, Capital One, Cloudflare, Microsoft, Wiz, Visa, Cisco, CrowdStrike, Akamai, Cognition, Perplexity, Uber, LangChain, and Veeam. Amazon and Visa joined the alliance; Amazon contributed Strands Agents and Cedar, while Visa contributed the Visa Vulnerability Agentic Harness. CrowdStrike fine-tuned NVIDIA Nemotron Nano for cyber defense.