FY27 National Defense Authorization Act (Chairman's Mark)

The FY27 National Defense Authorization Act is the annual U.S. defense authorization bill. On May 26, 2026, the House Armed Services Committee released the Chairman's mark, which Inside AI Policy described as "replete with AI provisions." The mark includes updates to autonomous-weapons rules, provisions accelerating deployment of AI models inside the Department of Defense, AI incident-and-vulnerability reporting requirements, and a directive for an AI-enabled manufacturing risk-response capability.

Status and timeline

On May 26, 2026, the House Armed Services Committee released the Chairman's mark of the FY27 National Defense Authorization Act and scheduled a full-committee markup for June 4, 2026. Inside AI Policy described the text as "replete with AI provisions" and characterized it as the most consequential AI legislative vehicle of 2026 to that date (Source: insideaipolicy.com).

As the NDAA was marked up, Sen. Mark Warner on June 11, 2026 backed Intelligence authorization provisions governing AI in targeting systems (Source: insideaipolicy.com). The same day, House defense appropriators proposed billions of dollars for AI and autonomous-warfare initiatives paired with expanded R&D funding flexibility (Source: insideaipolicy.com).

Sen. Adam Schiff (D-CA) proposed an amendment, disclosed July 9, 2026, that would let entities share information and coordinate to delay AI model releases without federal antitrust liability, provided they first disclose their plans to the relevant agencies; cloture was filed on a motion to proceed to S.4784 when the Senate returned July 13 (Source: insideaipolicy.com). See Adam Schiff.

On July 13, 2026, the Senate prepared to begin floor consideration of the NDAA, with bipartisan proposals limiting China's access to advanced computing expected to dominate the bill's AI debate (Source: insideaipolicy.com). The House passed its version of the bill 216-212, advancing its AI provisions toward Senate negotiation, per July 23, 2026 reporting (Source: insideaipolicy.com). On the House side the same day, the House Armed Services Committee directed the Pentagon to formalize department-wide procedures for blacklisting AI companies from Defense Department systems beyond the Chinese firms already barred — a provision reported months after the administration's dispute with Anthropic — and separately pressed it to examine AI-enabled manufacturing as a response to missile-production bottlenecks (Source: insideaipolicy.com; insideaipolicy.com). See DOD — Department of Defense (AI Deployer) and Export Controls (AI).

Sen. Mark Warner (D-VA) proposed an amendment, reported July 14, 2026, that would require federal agencies to remove AI models that fail criteria on child sexual abuse material and non-consensual intimate imagery, a requirement with implications for open-source developers and for xAI (Source: insideaipolicy.com).

On July 14, 2026, Senate negotiators placed three chip- and AI-related bills in the NDAA manager's amendment: the AI Overwatch Act, the MATCH Act, and the Chip Security Act (Source: punchbowl.news).

The Senate Armed Services Committee approved its version of the FY27 NDAA 18–9 on June 11, 2026. The Senate bill would authorize the Pentagon to create a four-star Robotic and Autonomous Systems Combatant Command — with test, evaluation, and limited acquisition authorities and operational guardrails, inspired partly by Ukraine's drone-focused service — and a new post, Under Secretary of Defense for Cyber, Information and Networks. The Pentagon was reported to be seeking roughly $55 billion for its Defense Autonomous Working Group, most of it through an unapproved reconciliation bill (Source: defenseone.com; insideaipolicy.com). See DOD — Department of Defense (AI Deployer).

AI provisions

The Chairman's mark, as described in the HASC release, contains several AI-related provisions.

The mark updates the rules governing autonomous weapons systems. This provision bears on Anthropic's public dispute with the Pentagon over zero-human-in-the-loop targeting and on Anthropic's stated "red line" against weapons that identify, track, and kill targets with zero human involvement.

Other provisions speed up the deployment of AI models across the Department of Defense, consistent with the Hegseth-era posture that "the risks of not moving fast enough outweigh the risks of imperfect alignment." See DOD — Department of Defense (AI Deployer).

The mark also establishes incident- and vulnerability-reporting requirements, creating new reporting obligations for AI incidents and AI-discovered vulnerabilities. These obligations parallel the Pentagon's posture on Mythos-class cyber-discovery (cf. Claude Mythos Preview).

A further provision directs the Pentagon to develop an AI-enabled manufacturing risk-response capability anchored to a planned network of dual-use factories, stating that "available digital and artificial intelligence (AI) capabilities can aid the Department in implementing the requirements of section 1841" of the FY26 NDAA. The provision ties two industrial-base threads together: the FY26 §1841 dual-use-factory network and the broader Pentagon AI-deployment posture, combined into a single tooling requirement (Source: insidedefense.com).

Relation to autonomous-weapons policy

According to Inside AI Policy, the mark is the first major U.S. statutory vehicle to attempt a comprehensive AI-and-defense framework rather than a one-off provision. If it survives full-committee markup and conference, it would set the operative U.S. legal posture on autonomous weapons while the Anthropic-DoD dispute reaches a D.C. Circuit decision (Source: insideaipolicy.com).

The autonomous-weapons update stands in tension with Anthropic's publicly stated use restrictions, and is consistent with Dario Amodei's separate written position that "fully autonomous weapons … may prove critical for our national defense." The accelerated-deployment provisions extend the DoDD 3000-09 framework into the frontier-model era.

Relationships

• depends-on: DoD Directive 3000.09 — Autonomy in Weapon Systems, Autonomous Weapons, Dual-Use Frontier AI.
• related: Anthropic v. United States (Pentagon ban challenge), DOD — Department of Defense (AI Deployer), Pete Hegseth, Dario Amodei, EO — Promoting Advanced AI Innovation and Security (Trump, signed June 2, 2026), Claude Mythos Preview.
• contradicts: Anthropic's stated red line against fully autonomous weapons (in the autonomous-weapons section, depending on final text).

Sources

• Inside AI Policy, May 27, 2026: "House Armed Services releases defense bill text replete with AI provisions" — (Source: insideaipolicy.com)