Senators add AI restrictions and prediction market bans to $1.1T Pentagon policy bill

The US Senate is embedding new guardrails on artificial intelligence and prediction markets directly into the Pentagon’s annual policy blueprint. Language added to the fiscal year 2027 National Defense Authorization Act, a bill worth roughly $1.1 trillion, would prohibit the military from using AI in nuclear weapon decision-making, restrict AI-driven surveillance of American citizens, and crack down on defense personnel trading on prediction markets with sensitive nonpublic information.

For the crypto-adjacent prediction market industry, that last part is particularly notable. The provision would require the Defense Secretary to establish regulations, complete with defined penalties, preventing military personnel and civilians from placing bets on platforms like Polymarket or Kalshi using classified intelligence.

What’s actually in the bill

The AI provisions trace back to legislation introduced by Senator Kirsten Gillibrand on June 2, 2026. Her framework establishes three core restrictions on how the Department of Defense can deploy artificial intelligence.

First, AI cannot be used in nuclear targeting or launch decisions. Human beings must remain in that loop, full stop. Second, AI-powered surveillance tools cannot be turned on US citizens. Third, autonomous weapons systems must maintain meaningful human oversight.

Senators Adam Schiff, Elissa Slotkin, and Mark Kelly have backed complementary measures calling for stricter controls on autonomous weapons.

The prediction market restrictions address a more specific problem. Allegations have surfaced involving military personnel engaging in what amounts to insider trading on prediction markets, reportedly tied to significant geopolitical events like the potential capture of Venezuelan leader Nicolás Maduro. The bill would task the Defense Secretary with building out enforcement mechanisms and specific penalties for violations.

The bigger picture on military AI

This isn’t the first time Congress has tried to put guardrails on Pentagon AI. The fiscal year 2026 NDAA also contained AI-related provisions, making this a continuation of a multi-year legislative trend rather than a one-off reaction.

Senator Gillibrand’s framework doesn’t ban military AI development or deployment broadly. It targets specific use cases where the consequences of algorithmic error, or algorithmic autonomy, are catastrophic and irreversible. Nuclear weapons sit at the top of that list for obvious reasons.

The surveillance restriction addresses a different concern entirely. Codifying the prohibition in the NDAA gives it the force of law rather than leaving it to executive branch discretion, which can shift with administrations.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.