US military uses AI tools in operations against Iran

US Central Command said US forces use AI tools to screen and organize data for operations against Iran, letting human analysts focus on verification and higher-level analysis, Capt. Timothy Hawkins said.

The US has hit more than 2,000 targets, including 1,000 within the first 24 hours of the campaign, according to reports.

People familiar with the operations told Bloomberg the military uses Palantir Technologies Inc.’s Maven Smart System.

Previous public US military statements said Maven is fed by more than 150 data sources.

Anonymous sources said Anthropic’s Claude large language model is installed on the system and is central to operations.

Capt. Hawkins said AI tools help generate points of interest and organize information but do not replace human decision-making, which follows a rigorous legal process.

The Stop Killer Robots coalition, representing 270 human-rights groups, warns AI decision-support systems risk automation bias and dangerously narrow the gap between recommending and executing strikes.

Centcom is investigating reports that a strike on a girls’ primary school killed more than 160 people; it is unclear who was responsible or whether AI played a role.

Palantir and Anthropic declined or did not respond to requests for comment.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

The Pentagon’s dispute with Anthropic focused on two ethical red lines

Frictions rose after Anthropic declined to drop two carve-outs it sought on lawful use of its AI model, Claude. The carve-outs covered mass domestic surveillance of Americans and fully autonomous weapons 1.
The Pentagon pushed for terms allowing “all lawful purposes.” When Anthropic did not budge, the Trump administration moved to label it a “supply chain risk.” That label has usually targeted U.S. adversaries, and it had not been publicly used on an American company before 1.
Fortune reported Anthropic held a $200 million Pentagon contract, and it has now been cancelled 2.
OpenAI later announced an agreement for the Pentagon to use its AI models in classified systems. Fortune reported terms tied to “any lawful purpose” language. OpenAI also said it uses similar limitations and guardrails, which suggests the split hinged partly on contract wording versus policy and technical controls 2.

The dispute points to platform-style mission AI and the business risks of government deals

Bloomberg sources said the military uses Palantir Technologies Inc.’s Maven Smart System. Anonymous sources said Anthropic’s Claude runs on that system and supports operations.
A common setup pairs a specialized data and software platform layer, such as Palantir’s Maven, with swappable AI models. The platform helps screen and organize data for military operations and keeps rules on data access, security, and human oversight 3.
This structure can let the Department of Defense work with multiple frontier model providers while keeping governance over deployment 4.
Anthropic’s experience warns tech firms about the downside of entering tightly regulated sectors, including defense.
Turning down a powerful buyer on ethical grounds can carry steep costs. The “supply chain risk” move raised fears it could chill enterprise deals tied to Pentagon work, while legal and policy experts questioned how broad and lawful a sweeping ban would be 2, 1.