OpenAI Hires Former Army Secretary Dan Driscoll as National Security Counselor

OpenAI has hired Dan Driscoll, the former US Army Secretary, as senior counselor for national security modernization, a role the company created for him. CNBC reported the hire on November 3, 2026, citing OpenAI and Driscoll's updated LinkedIn profile.

No contract, budget or reporting line came with the announcement. This is a people move that shows where OpenAI wants to grow: government and allied national-security customers.

Agents that act on your behalf need rules. See how explainx helps secure your agents →

TL;DR: facts and open questions

What OpenAI and Driscoll said

According to CNBC, Driscoll will advise OpenAI's leadership on AI applications in national security and cybersecurity and work with senior leaders and external partners. A particular focus is helping US allies adopt advanced AI responsibly.

Sam Altman said the United States has "an extraordinary opportunity to work with other countries to spread the adoption of democratic AI," and that Driscoll's appointment would help strengthen American leadership in the technology.

Driscoll said: "National security requirements are changing faster than at any time in the past fifty years." He added that AI could help reduce paperwork, process information and deliver answers more quickly to people working in national security. In his own words on the announcement, he is "joining OpenAI to help the U.S. and our allies put this technology to work responsibly, for the people who keep us safe."

Who Driscoll is

Driscoll was confirmed as Army secretary by a 66-28 Senate vote in February 2025, DefenseScoop reported. He is an Army veteran who deployed with the 10th Mountain Division to Iraq, worked in venture capital and private equity, and was a senior adviser to JD Vance before Vance became vice president.

In confirmation answers he argued the Army should buy more commercial off-the-shelf technology, listing small drones, counter-drone systems, electronic warfare gear and communications equipment. CNBC adds that he pushed modernization, including wider access to drone and counter-drone technology among US allies, and that he met Ukrainian President Volodymyr Zelenskyy in Kyiv in November 2025 about a US-proposed peace plan.

CNBC says he stepped down as Army secretary earlier this year after months of reported tensions with Defense Secretary Pete Hegseth, citing Reuters. We did not read the underlying Reuters report ourselves, so we repeat it only as CNBC's attribution.

Why this hire matters

1. A buyer's translator

Defense and intelligence agencies buy technology through rules and habits that AI labs rarely learn from the outside. A former service secretary who argued for commercial off-the-shelf purchasing is a person who knows how to explain a model to a procurement office. The job description, helping teams "understand the operational needs of global national security institutions," says as much.

2. "Democratic AI" as a sales frame

Altman's phrase, "democratic AI," is the pitch to allied governments: US models and infrastructure as an alternative to Chinese ones. The hire pairs that pitch with a person who has worked directly with allies on drones and counter-drone systems. That is consistent with other OpenAI moves in the Gulf, such as the ADNOC and G42 Frontier deal.

3. Cyber is the bridge

OpenAI's Daybreak work is the existing national-security-adjacent product line. We covered the launch in OpenAI Daybreak and Codex Security, the September follow-up on the OpenAI defense factory, and recent bounty and patching programs in OpenAI's bug bounty and security sprint. A national-security counselor sits naturally on top of that: cyber defense is the use case governments will pay for first.

4. Competition over government work

AI labs are competing for public-sector trust at the same moment that the relationship between labs and the Pentagon is contested. Our post on the DC Circuit decision on Anthropic's Pentagon blacklist shows how fast that relationship can change, and our report on protesters interrupting Hegseth's autonomous warfare panel shows the public pushback around military AI. Hiring someone with Army credentials and a documented record on drones is a bet that OpenAI can stay on the right side of both sets of facts.

Timeline: from Army secretary to OpenAI

Ethics rules on post-government employment typically restrict lobbying and certain contacts with former agencies, and a counselor role that "works with government officials" will be watched against those rules. Neither OpenAI nor Driscoll has addressed that in what we read.

Questions people are asking

Is this a political hire? DefenseScoop describes Driscoll as a former senior adviser to JD Vance. That helps with access and also draws scrutiny. OpenAI has framed the role as practical: advising on use cases and allied adoption.

Will OpenAI models go into weapons? Nothing in the announcement says so. The stated focus is national security modernization and cybersecurity, with Daybreak named as the base. OpenAI's usage policy and any contract terms would answer the weapons question, not a job title.

Why would an AI lab need a former Army secretary? Because the buyer is unusual. Procurement cycles, security clearances, accreditation and operational doctrine shape what a defense customer can adopt. An executive who has run a large service can shorten the learning curve.

How does this compare with rivals? Other labs already work with defense agencies and have had their own disputes, as the Anthropic blacklist case shows. We have not seen comparable hires announced today, and we do not rank labs by defense readiness here.

What should allies expect? The stated plan is to help US allies adopt advanced AI. In practice that probably means briefings, pilot programs and partnerships on cyber defense and administrative workloads first, because those are the least controversial uses. The announcement does not name a country or program.

The tension: military credentials versus public trust

A few things deserve plain statement.

• Revolving door. Hiring a former official soon after he left office is common in defense and technology. It also invites questions about access and influence. We have no evidence of wrongdoing here; the point is that readers and regulators will ask.
• Dual-use. Cyber-defense tools can also help offense. OpenAI has discussed these risks in its own cyber work, and its preparedness decisions are covered in our post on Astra cybersecurity and the Preparedness Framework.
• Security of the company itself. OpenAI has been warning staff about counterintelligence risks, as in OpenAI's honeypot spy warning. A closer relationship with national-security institutions raises the stakes of that work.
• Military users and safety limits. How OpenAI's usage rules apply to military uses is a recurring question. The appointment does not answer it.

Claimed vs verified

Different outlets give slightly different timing for his resignation. We rely on CNBC's "earlier this year" and do not state a date.

What to watch next

1. Whether OpenAI announces new government or allied partnerships that cite the counselor role.
2. Whether OpenAI publishes terms for national-security use of its models.
3. Reactions from Congress and watchdogs about the hire.
4. Whether the role expands into a formal national-security team under a named executive.

What this means for people who build with OpenAI

For most developers, nothing changes today. If you build for government customers, expect OpenAI sales and compliance conversations to become more fluent in defense procurement. If you worry about military uses of general models, read OpenAI's usage policies and watch for any change tied to this role. If you run agents that touch sensitive systems, runtime controls matter regardless of who the vendor is. AgentBeam is the agent security platform from the explainx.ai team that stops AI agents before they take dangerous actions.

Related reading

• OpenAI Daybreak and Codex Security
• OpenAI defense factory: agent-first cyber defense
• OpenAI bug bounty and security sprint
• DC Circuit and the Anthropic Pentagon blacklist
• Hegseth's autonomous warfare panel and protesters
• OpenAI honeypot spy warning to staff
• ADNOC, OpenAI and G42 Frontier

Details are accurate as of November 3, 2026.