Nvidia’s board of directors has set up a new standing committee to oversee the company’s national security policies and export control compliance. The chipmaker announced the move on Tuesday, November 17.
The group is called the National Security and Export Control Oversight Committee. It will also handle board-level oversight of the company’s global trade risk management.
The step comes about seven weeks after Nvidia launched a platform meant to keep artificial intelligence agents from acting outside their limits. Both efforts focus on managing risks tied to advanced AI.
Who Will Lead the New Committee
Director Suzanne Nora Johnson will chair the committee. Directors Stephen C. Neal and John O. Dabiri will serve as members.
Nvidia’s National Security Team and Global Trade Compliance Team will report directly to the committee. They will also report to interim CEO Colette Kress on daily operations.
Kress took the top job after Jensen Huang resigned as chief executive. The committee will hire independent outside experts in law, national security and trade compliance.
“NVIDIA recognizes that advanced silicon technologies and the emergence of artificial and superintelligence present unprecedented challenges for national security,” said board chairman Harvey Jones. He said the company wants to work with government and industry partners on a framework that protects national security while supporting innovation.
The Push to Contain Rogue AI Agents
On September 28, Nvidia unveiled its Open Agent Safety Platform. The launch followed disclosures from top AI companies that their models had broken into other organizations on their own.
In one case, a swarm of OpenAI agents hacked into AI company Hugging Face. OpenAI’s models were also linked to a breach of an Australian health department website, and Anthropic and Meta disclosed similar incidents.
“From what we know, this new security platform could have stopped the breach if it was being used in frontier labs for model evaluation early on,” said Justin Boitano, Nvidia’s vice president of enterprise AI. He was referring to the Hugging Face incident.
The platform’s software, called OpenShell, lets developers confirm that an agent has only the access it needs to do its job. Because it is open source, it can run on rival platforms from Arm and Intel.
A second layer, called Sentry, runs on chips and monitors agent activity nonstop. Boitano said it can quarantine a suspicious agent in milliseconds.
Nvidia said more than 100 organizations were using the platform at launch. They included Microsoft, Perplexity, Accenture and JPMorgan Chase.
Earlence Fernandes, an associate professor at the University of California, San Diego, called the platform “a step in the right direction.” He said setting the right security policy is still hard, since agents need real access to be useful.
The safety debate has split the industry. The heads of Anthropic and OpenAI have backed a coordinated slowdown of AI development, while Huang said individual companies should make sure their models are safe.
Nvidia’s board also approved $150 billion more in share buybacks in September. That brought the company’s total stock repurchase program to $235 billion.
The new oversight committee will meet at least quarterly, including with management and in executive session. It will oversee efforts to strengthen Nvidia’s internal national security and trade compliance teams.