The Open Secure AI Alliance, now comprising over 120 organizations, is developing new guidelines called Shared AI Findings Exchange (SAFE) to enhance cybersecurity for agentic AI. Spearheaded by members including NVIDIA, Cisco, CrowdStrike, Hugging Face, and Red Hat, and facilitated by the Linux Foundation, SAFE aims to transform AI cybersecurity incidents into shared defenses for the entire ecosystem. The proposed guidelines focus on confidential analysis of AI incidents, informing affected parties, identifying recurring failures, and publishing recommendations to mitigate systemic risk. NVIDIA is also contributing a suite of open cybersecurity software and models, such as the NOOA research harness and the Garak LLM vulnerability scanner, to bolster AI security across various layers of the technology stack. AI
IMPACT Establishes a framework for collective defense against AI-driven cyber threats, potentially accelerating secure AI development.
RANK_REASON Industry alliance proposes new cybersecurity guidelines for AI agents, involving major tech companies and a foundation.
- Black Hat
- Cisco
- CrowdStrike
- Hugging Face
- Linux Foundation
- NVIDIA
- NVIDIA Alpamayo
- NVIDIA BioNeMo
- NVIDIA Cosmos
- NVIDIA Isaac GR00T
- NVIDIA Labs Object-Oriented Agent
- NVIDIA Nemotron
- Open Secure AI Alliance
- Red Hat
- SAFE
- Anthropic PBC
- Claude
- NOOA
- Volta Infra Holdings Ltd.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
