Just one week after its formation, the Open Secure AI Alliance (OSAA), an industry consortium spearheaded by Nvidia, has already established a working group focused on AI security. This group, named the Shared AI Findings Exchange (SAFE), is quickly moving to present initial proposals for public review. The Linux Foundation is facilitating the management of these proposals.

The SAFE group developed its initial guidelines during the Black Hat conference in Las Vegas, a prominent event in the cybersecurity community. The current proposals address fundamental aspects of AI incident response, including confidential reporting of cybersecurity breaches, timely notification of affected parties, and conducting blame-free analyses to foster collective learning within the industry.

Beyond establishing security protocols, OSAA members are actively contributing open-source technologies intended to bolster AI security. The long-term vision for this initiative is to create an open-source ecosystem that enables enterprises to secure their AI agents and defend against potential threats, such as malicious AI actors. Companies like Nvidia are contributing tools such as the Garak LLM vulnerability scanner, while Okta is developing agent identity technology, Red Hat is focusing on agent governance, and Amazon has provided an agent building tool and an authorization language.

The alliance has rapidly attracted over 120 companies, including major players like Adobe, BlackRock, Cisco, Intel, Microsoft, and Visa. However, notable absences include Anthropic, OpenAI, and Google, despite both OpenAI and Google having previously signed an open letter advocating for open-source AI development and support from the White House. The swift progress of the OSAA is seen by some as a significant positive development for the U.S. open AI ecosystem, potentially offering a robust path toward ensuring AI safety and security.

Key Takeaways

• The Open Secure AI Alliance (OSAA), launched by Nvidia, has quickly formed a working group focused on AI security protocols.
• The SAFE working group is proposing guidelines for reporting, alerting, and analyzing AI cybersecurity incidents.
• OSAA members are contributing open-source tools and technologies to build a shared ecosystem for AI security, though key players like OpenAI and Google are currently absent.

Editor’s Analysis & Impact

The rapid mobilization of the OSAA underscores the urgency the tech industry places on AI security and open-source collaboration. By establishing the SAFE working group and soliciting public comment on initial proposals, Nvidia and its partners are demonstrating a proactive approach to addressing emerging AI threats. The contribution of open-source tools signals a commitment to building a shared defense infrastructure. While the absence of major players like OpenAI and Google is notable, their past support for open-source AI suggests potential future engagement. This initiative could significantly shape the future landscape of AI governance and security, fostering a more resilient and collaborative ecosystem.

Frequently Asked Questions

Q: What is the Open Secure AI Alliance (OSAA)?
A: The OSAA is an industry consortium, spearheaded by Nvidia, that aims to foster collaboration and develop security standards for open-source AI technologies. It has rapidly grown to include over 120 companies.

Q: What is the SAFE working group?
A: SAFE, or the Shared AI Findings Exchange, is a working group within the OSAA. Its primary focus is to develop proposals for how to confidentially report AI cybersecurity incidents, alert affected parties, and conduct blame-free analysis for industry-wide learning.

Q: Which major companies are part of the OSAA?
A: Current members include Adobe, BlackRock, Cisco, Intel, Microsoft, and Visa, alongside Nvidia. Notably absent are Anthropic, OpenAI, and Google.