On Tuesday, senior US military leaders including Defense Secretary Pete Hegseth met with executives of artificial intelligence company Anthropic as the simmering debate over AI’s military application reached a crucial phase.
The question is whether the US Government should have any restrictions on using Anthropic’s large-language model, Claude in highly sensitive defence operations.
Pentagon sent an ultimatum Anthropic demanding the company to accept their terms before Friday or else face severe consequences.
What is the Pentagon’s ultimatum to Anthropic?
The New York Times cites a senior Pentagon representative as saying that the Trump Administration has informed Anthropic it may invoke the Defense Production Act, if they do not meet the deadline of 5:01 pm on Friday.
This would force the company to make its AI available for military purposes.
Anthropic has been branded a “supply chain risk” by officials, a label usually reserved for firms linked with foreign enemies.
This could prevent the US Government from ever using Anthropic products.
Both measures have fundamentally different goals.
The first would require the military to adopt the Anthropic model while the second would ban its use.
The contradictory threat reflects both frustration at Anthropic’s stubbornness and its strategic importance.
An Anthropic representative said that the meeting on Tuesday “continued to have good faith conversations about our use policy in order to ensure Anthropic could continue to support government national security missions, consistent with what models are capable of reliably and responsible doing,” Reuters reported.
Anthropic is important to military personnel
Anthropic holds a unique place in the US Defense ecosystem.
The AI model of this company is the only one that is currently operating on classified systems.
Last July, the Department of Defense entered into contracts with a number of leading AI companies — such as Google and OpenAI — for deals up to 200 million dollars.
Until last week, the only model approved for classified environments was Anthropic’s Claude.
AI and the core of disagreement
Anthropic is known as one of the leading AI companies that puts safety first. This has led to increasing disagreements with the defence department.
US military leaders are pushing for a broader, less restrictive access to Claude’s capabilities.
According to those familiar with the discussions, Anthropic has refused to allow its models to become part of mass surveillance systems or autonomous weaponry that can make deadly decisions without human intervention.
This dispute escalated in the first week of this month when Pentagon officials were concerned Anthropic was asking questions regarding how their AI tools had been used during an operation that captured President Nicolas Maduro.
Claude has been integrated into the Department of Defense’s workflow, but it has also threatened to cut ties with Claude over artificial restrictions imposed by a contractor.
NYT reports that Pentagon officials claim it is not the responsibility of vendors to dictate how software or weapons should be used. Instead, they argue, this is the responsibility of government.
Anthropic supporters argue, however, that Anthropic is being punished because it was the first company to enter this market and developed a model aimed at government, called Claude Gov. This product differs significantly from Anthropic’s public facing products.
The next step: Legal and commercial implications
According to Reuters a source familiar with this matter stated that Anthropic had no plans of easing restrictions on its use for military purposes even though discussions are ongoing with the Pentagon.
Reuters reported also that Anthropic’s chief executive Dario Amedei informed Hegseth at the meeting on Tuesday that his company hadn’t raised any concerns about the raid with either the Pentagon, or the defense contractor Palantir.
Amodei said that the current safeguards will not affect the existing defense department operations.
Anthropic may suffer consequences that go beyond its defense contracts if the Pentagon decides to designate supply-chain risks.
A label like this could damage the relationships of the firm with firms who do business with US Government.
Franklin Turner is a lawyer for McCarter &English who specializes in government contracting. Reuters cited his comments.
The unusual nature of his threats could lead to extensive litigation.
Anthropic claims that for the moment, talks continue in good faith.
If the company is successful in maintaining its safety-first approach while still being a major supplier of the US Military, it will determine its future and how AI technology is managed within national security environments.
The ICD: Why is the Pentagon at war with Anthropic? The ICD first published this post
This site is for entertainment only. Click here to read more
