The newly formed Open Secure AI Alliance is already developing new guidelines to strengthen AI cybersecurity against agentic AI attacks, to be presented during Black Hat USA 2026 today. The alliance has been backing the development of more open-source models in the industry. Nvidia CEO Jensen Huang (in his first post on X) made clear that “Open models strengthen safety and cybersecurity” before the alliance was even made public.

A week has passed since Nvidia united with Microsoft, Linux, and many more to create the Open Secure AI Alliance. The initiative was put together not long after a rogue OpenAI agent escaped its sandbox testing environment and went on to hack into popular AI platform Hugging Face. Powered by two closed-source OpenAI models, it worked autonomously after its initial setup. The co-founder of Hugging Face went on to say these agents will become the “most common” type of cyberattack.

SAFE guidelines are being proposed at Black Hat USA 2026 today

The Open Source AI Alliance is now more than 120 organizations strong. When it was set up last week, there were a total of 37 members, including Nvidia, Hugging Face, Microsoft, and The Linux Foundation. OpenAI was notably not present in the list. The initial proposal for the Shared AI Findings Exchange (SAFE) guidelines is being worked on by The Linux Foundation alongside other members; the foundation set up a Request for Comments (RFC) on GitHub earlier today.

The SAFE guidelines are being drafted by an Open Secure AI Alliance working group. NVIDIA, Cisco, CrowdStrike, Hugging Face and Red Hat are among Open Secure AI Alliance members working with the Linux Foundation to contribute to the initial proposal.

Source: Nvidia Company Blog

The draft RFC proposes a framework built around several core principles:

• Confidential reporting of AI security incidents and near misses
• Timely notification of affected organizations
• Collaborative analysis focused on learning rather than blame
• Structured reviews spanning the complete AI operating stack, including models, safeguards, tools, runtime environments, monitoring, human operations, and supply chain dependencies
• Evidence-based operational guidance that organizations can implement and verify
• Independent governance representing stakeholders from across the AI ecosystem

Source: The Linux Foundation blog

Black Hat USA 2026 is a cybersecurity event which has been taking place since the start of August, but today (August 4) marks the ‘Summit Day’ featuring technical deep-dives, including the AI Summit. Black Hat says “AI now sits at the core of modern security” while, on the flip side, “enabling faster, more sophisticated attacks” – such as the one at the hands of OpenAI’s rogue agent.

About the Author

Jack Goodall

At PC Guide, Jack is mostly responsible for reporting on hardware deals. He also specializes in monitors, TVs, and headsets and can be found putting his findings together in a review or best-of guide.