# OpenAI’s Dean Ball warns of government monopoly risks in AI oversight > Source: > Published: 2026-06-22 06:15:24+00:00 # OpenAI’s Dean Ball warns of government monopoly risks in AI oversight Former Trump administration advisor turned OpenAI policy lead says classifying frontier AI testing could concentrate dangerous power in intelligence agencies Dean Ball has a message for Washington: the government’s growing appetite for controlling who gets to test the most powerful AI systems is a fast track to a monopoly nobody voted for. Ball, who joined OpenAI in July 2026 to lead its new strategic futures team, laid out his concerns in a recent podcast interview. His core argument is that classifying frontier AI testing procedures, a move the Trump administration has been pursuing, risks locking advanced models behind government gates and blurring the lines between regulatory oversight and corporate operations. ## From the White House to OpenAI Ball’s trajectory makes his warnings hard to dismiss. He previously served as a Senior Policy Advisor in the Trump administration, where he was instrumental in shaping America’s AI Action Plan. Before that, he held posts at the Foundation for American Innovation and was a visiting fellow at the Heritage Foundation. His new role at OpenAI, announced around mid-June, puts him at the helm of a team focused on frontier AI policy. The portfolio covers some of the thorniest questions in the field: catastrophic risk management, recursive self-improvement of AI systems, and the broader societal implications of these technologies as they interact with government institutions and labor markets. ## $58.5 billion and the sovereign AI arsenal Ball didn’t just deal in abstract warnings. He pointed to a very concrete number: the Department of War’s fiscal year 2027 budget request includes $58.5 billion for AI-related initiatives. Of that, $46 billion is earmarked for what’s being called a “sovereign AI arsenal.” Ball specifically flagged the concentration of oversight powers within intelligence agencies like the NSA. His argument is straightforward: intelligence agencies are built for secrecy, not governance. Handing them the keys to AI oversight means decisions about who gets access to frontier models, and under what conditions, get made by organizations that lack deep AI expertise and face zero public accountability. He also noted a telling development in the contracting landscape. According to Ball, the Department of War has been winding down its contracts with Anthropic while continuing relationships with other AI companies. ## Private governance over public regulation Ball has long advocated for private governance mechanisms over stringent public regulations in the AI space. He was a prominent critic of California’s SB 1047, the state-level AI safety bill that would have imposed liability frameworks on developers of frontier models. Ball pushed back against that approach, arguing it would chill innovation without meaningfully improving safety outcomes. Instead, he’s championed what amounts to a market-driven safety model: independent verification of AI systems, economic diffusion of the technology across a broad base of actors, and governance frameworks that keep the private sector in the driver’s seat while maintaining transparency. ## What this means for investors The $58.5 billion budget figure alone should have anyone in the AI investment space paying attention. Ball’s warnings highlight that level of government involvement, combined with classification of testing procedures, could effectively create a two-tier AI market. The tension between Anthropic’s apparent loss of Department of War contracts and the continuation of others also suggests that government procurement in AI is becoming more politically and strategically selective. Investors tracking the AI defense sector should watch not just who wins contracts, but who loses them, and whether any pattern emerges in how those decisions correlate with companies’ stances on safety, openness, and government cooperation. **Disclosure:** This article was edited by Editorial Team. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).