“Hegsseth’s demand would in principle extend to apply Anthropic’s software to nuclear weapons without humans in the loop.”
Our reading, the sentence that was checked:
Hegseth's demand would in principle allow Anthropic's software to be applied to nuclear weapons without humans in the loop.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, America, and probably the world, stands on a precipice., paragraph 13.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-america-and-probably-the-world-stands-db7386-a1
We recorded 9 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-02-26. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- in principle (qualifier, stated by the speaker)
From the speaker
- Hegseth's demand
- would in principle extend
- to apply Anthropic's software to nuclear weapons
- without humans in the loop
Our interpretation
- Read as a claim about the scope of the demanded terms: did the 'any lawful use' terms, with the autonomous-weapons safeguard removed, contain any exclusion that would keep Claude from being used in nuclear weapons decisions without a human in the loop?
- Whether a binding legal limit already excluded such use matters. If 'any lawful use' already excludes it under US law or policy, the demand would not extend to it even in principle.
Counts as supported if: The demanded terms had no nuclear exclusion and no human-in-the-loop requirement, and no binding law forbids AI-controlled nuclear employment, so 'any lawful use' would cover it.
Counts as contradicted if: The demanded terms, or binding law that 'lawful use' incorporates, required human control over nuclear employment decisions.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The demand was 'any lawful use', with Anthropic's safeguard against fully autonomous weapons removed. No source shows a nuclear carve-out in the terms, which fits Marcus's hedged 'in principle' reading. But a senior Pentagon official said its policy keeps a human in the loop on all nuclear employment decisions. Section 1638 of the FY2025 defense law also states a US policy requiring human action in those decisions. Whether those rules would make AI-controlled nuclear use 'unlawful', and so outside the demand, depends on legal readings the record does not settle. The contract text is not public either.
Limitations
- Rests on interpretation of 'lawful' and 'in principle'.
- The Arms Control Association page is a secondary summary of the statute.
Evidence (3 items)
Supports the claim: Anthropic,
The Department of War has stated they will only contract with AI companies who accede to “any lawful use” and remove safeguards in the cases mentioned above.
Why it matters here: The demand was bounded only by 'lawful use', and the fully autonomous weapons safeguard was to be removed. This is consistent with Marcus's inference, but it is not explicit about nuclear weapons.
- Publisher
- Anthropic
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Goes against the claim: Washington Post (via Anchorage Daily News),
“It remains the Department’s policy that there is a human in the loop on all decisions on whether to employ nuclear weapons,” a senior defense official said.
Why it matters here: Pentagon policy kept a human in the loop for nuclear employment, which cuts against the inference. But a policy is not a contract term, and it can change.
- Publisher
- Washington Post (via Anchorage Daily News)
- Kind of source
- News report
- Describes events of
- Public from
Goes against the claim: Arms Control Association,
“It is the policy of the United States that the use of artificial intelligence efforts should not compromise … the principle of requiring positive human actions in execution of decisions by the President with respect to the employment of nuclear weapons.”
Why it matters here: Statute declares a US policy of human action in nuclear employment decisions. Whether this binds 'lawful use' is unclear, since it is a policy statement rather than a prohibition.
- Publisher
- Arms Control Association
- Kind of source
- Secondary summary
- Describes events of
- Public from
What would settle it
Publication of the Pentagon's proposed contract language, or an authoritative legal reading of whether 'any lawful use' excludes autonomous nuclear employment under Sec. 1638 or binding DoD rules.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, America, and probably the world, stands on a precipice.
Hegsseth’s demand would in principle extend to apply Anthropic’s software to nuclear weapons without humans in the loop.
- Where
- paragraph 13
- Published
- 2026-02-26
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-america-and-probably-the-world-stands-db7386-a1: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-america-and-probably-the-world-stands-db7386-a1.