AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › People & Organizations

Emil Michael

medium confidence · updated 2026-09-27

Under Secretary of Defense for Research and Engineering (confirmed May 2025), the Pentagon's chief technology officer; former Uber chief business officer. Counterparty to Dario Amodei in the 2026 Anthropic–Pentagon dispute over safety guardrails, and the official under whom CDAO was realigned in the August 2025 reorganization.

Contents

Emil Michael is the Under Secretary of Defense for Research and Engineering (USD(R&E)), the Pentagon's chief technology officer. The Senate confirmed him on May 14, 2025 by a 54–43 vote, and he was sworn in on May 20, 2025; following the department's secondary redesignation, his office is also styled Under Secretary of War (Research and Engineering) (Source: defensescoop.com; Source: cto.mil; Source: war.gov).

Background

Michael was chief business officer at Uber from 2013 to 2017, where his official Defense Department biography credits him with creating global partnerships and driving over $20 billion in capital-raising initiatives (Source: war.gov). He was nominated for the USD(R&E) post in December 2024 and confirmed in May 2025 (Source: breakingdefense.com). As USD(R&E) he serves as the Defense Department's point person on advancing technology, including its AI portfolio; Sen. Elizabeth Warren, ahead of his confirmation, sought ethics commitments given the role's authority as the department's chief technology officer (Source: warren.senate.gov).

Role in Pentagon AI organization

In August 2025, a memo by Deputy Secretary Stephen Feinberg moved the Chief Digital and Artificial Intelligence Office (CDAO) under USD(R&E), placing the Pentagon's central AI office in Michael's reporting line — a change the Congressional Research Service characterized as an effective demotion of CDAO (see Chief Digital and AI Office (CDAO)).

Dispute with Anthropic

Michael was the principal Defense Department counterparty in the 2026 dispute between Anthropic and the Pentagon over safety guardrails on military AI use. Court documents released July 2, 2026 in one of Anthropic's lawsuits against the Pentagon (Anthropic v. United States (Pentagon ban challenge)) revealed months of correspondence between Anthropic CEO Dario Amodei and Michael over guardrails for the Pentagon's use of AI, with tensions persisting even as Anthropic resolved its separate export-control dispute with the administration on June 30, 2026 (Source: wsj.com).

The August 27, 2026 summary-judgment opinion in that case places a four-page memorandum by Michael, dated March 2, 2026, at the centre of the government's case, describing it as supplying "the entirety of the government's rationale" for the supply-chain-risk designation. The memorandum asserted that AI models "are acutely vulnerable to manipulation," that Anthropic could "attempt to disable" or "alter the behavior of the model . . . in the middle of ongoing warfighting operations, if it feels that its 'redlines' are crossed," and that the Department "would be forced to operate a black box controlled by a hostile party, which could contain hidden biases or backdoors." Its conclusion attributed the escalation to Anthropic's conduct in the negotiations and to the company "engaging in an increasingly hostile manner through the press" (Anthropic PBC v. U.S. Department of War — Order on Cross Motions for Summary Judgment (N.D. Cal., August 27, 2026)).

The court made three findings about Michael's role. First, on the substance: the government abandoned the backdoor-access theory by the summary-judgment stage, and the administrative record contains nothing describing "what technological means would give rise to the so-called 'backdoors.'" Second, on the procedure: 48 C.F.R. § 239.7304(a) requires the joint recommendation to rest on a risk assessment by the Under Secretary of Defense for Intelligence, and the assessment instead came from Michael, "the same person who was simultaneously engaged in contract negotiations with Anthropic." Third, on pretext: on the morning of March 4, 2026, the day after the designation was finalized, Michael emailed Anthropic saying "After reviewing with our attorneys and seeing your last draft (thanks for being fast), I think we are very close here," and that he "hopes this works as I am running out of time." The court admitted that email as evidence of agency bad faith, calling it "wholly inconsistent" with his determination two days earlier that Anthropic was a hostile company presenting an "unacceptable national security threat." It also noted that his March 17, 2026 declaration asserted it was "reasonably likely" Anthropic would alter or shut off the Department's use of Claude during an operation, while his March 24 declaration dropped the claim entirely (Anthropic PBC v. U.S. Department of War — Order on Cross Motions for Summary Judgment (N.D. Cal., August 27, 2026)).

The D.C. Circuit, reviewing the parallel exclusion under 41 U.S.C. § 4713, gave Michael's memorandum more weight. It described Hegseth's March 3, 2026 determination as resting on a recommendation from senior officials that in turn rested on Michael's memorandum, which cited Anthropic's refusal to allow all lawful uses, its ability to "alter system guardrails and model weights," and its questioning of the Department's use of Claude in a sensitive military mission abroad. The court quoted his account that an Anthropic executive's question about that mission "led to alarm by the DoW and the prime contractor who provides Anthropic software, and raised material doubts as to whether they would cause their software to stop working," and credited his conclusion that Anthropic might manipulate the "design, integrity, and operation" of the Department's Claude models, potentially causing "critical defense system[s] failing to engage." It accepted Anthropic's point that one sentence of the memorandum, suggesting Anthropic could disable its technology "in the middle of ongoing warfighting operations," was mistaken for models already delivered, but held that "the basic thrust of Michael's analysis remains intact" because Anthropic can alter guardrails in each new model it delivers (Anthropic PBC v. U.S. Department of War — D.C. Circuit Opinion, Nos. 26-1049 and 26-1162 (September 25, 2026)).

Relationships

Categories: People & Organizations | us | government | national-security | defense
What links here (8)