The Pentágono is keeping open the possibility of new talks with the Anthropic before the deadline on Friday to ease restrictions on the use of its artificial intelligence technology, a senior official from the Departamento of Defesa of the The statement comes amid a prolonged impasse over ethical safeguards in the Claude model, with the government demanding unfettered access for legal military purposes and the company resisting certain applications. The undersecretary of Defesa for Pesquisa and Engenharia, Emil Michael, stated that dialogue remains possible despite recent tensions.
The dispute involves a contract worth up to $200 million signed in recent years between Pentágono and leading AI labs, including Anthropic, OpenAI and Google, to integrate advanced models into classified systems. Departamento of Defesa seeks to remove limitations imposed by Anthropic that prevent the use of Claude in mass surveillance of American citizens or autonomous weapons. The company maintains these red lines to avoid risks to democratic values and uses that could compromise public safety.
The deadline set by the secretary of Defesa Pete Hegseth expires at 5:01 pm (ET) this Friday. Caso there is no agreement, the Pentágono can terminate the partnership, classify the Anthropic as a risk in the supply chain — a measure generally applied to entities linked to foreign adversaries — or invoke the Lei of Produção of Spokesperson Sean Parnell reinforced that the government will not accept private companies dictating military operational terms.
Anthropic’s position on safeguards
Anthropic reiterated its refusal to give in to Pentágono’s latest demands. CEO Dario Amodei has stated that the company cannot in good conscience accept terms that do not adequately address concerns about mass surveillance or fully autonomous weapons.
The startup argues that the proposals received overnight did not significantly advance the requested guarantees. Apesar of steadfastness, Amodei expressed a preference to continue serving Departamento of
Statements from Departamento officers of Defesa
Emil Michael highlighted this Friday morning that Pentágono remains willing to dialogue with Anthropic, even classifying the company’s behavior as unpredictable during the confrontation over AI safeguards.
The undersecretary emphasized the importance of continuing negotiations before the deadline expires. Autoridades of
Context of recent negotiations
The clash gained intensity after a meeting between Hegseth and Amodei at the beginning of the week. The secretary imposed an ultimatum to sign a document that allows full use for legal purposes, under threat of drastic measures.
Pentágono consulted defense contractors about dependence on services from Anthropic, indicating preparation for possible disruption. The company currently operates on classified networks from Departamento to Defesa, which makes its technology strategic for military operations.
Potential impact on military AI contract
The deal under discussion is part of a series of $200 million cap contracts signed with leading AI labs over the past 12 months. Esses arrangements aim to accelerate the adoption of advanced tools in classified environments and defense operations.
Refusal to remove restrictions may lead to Anthropic being excluded from future government opportunities. Outras companies in the sector have already accepted similar terms, allowing broad use subject only to American law.
Chronology of the impasse
Negotiations have been dragging on for months, with a focus on balancing technological innovation and ethical controls. Anthropic has positioned itself from the beginning as an advocate of robust safeguards in its models.
The confrontation escalated with the sending of contractual language considered final by Pentágono. The company’s response came in an official post on the website, reinforcing the impossibility of accepting the current conditions.
Friday’s deadline represents a critical point for the resolution or definitive break of the partnership.