Nature Investigation Reveals Growing AI Use in Warfare as US-Iran Conflict Escalates
Key Takeaways
- ▸The US military's Maven Smart System, using AI for target prioritization, has been deployed in the February 2026 US-Israeli attacks on Iran
- ▸Anthropic lost its $200 million Department of War contract after refusing to remove safeguards preventing its Claude AI from guiding autonomous weapons or conducting mass surveillance
- ▸OpenAI replaced Anthropic as the government's AI supplier, agreeing to restrictions on autonomous weapons despite broader government directives
Summary
A Nature investigation has highlighted the expanding role of artificial intelligence in modern warfare, particularly during the escalating US-Israeli-Iran conflict that began on February 28, 2026. The US military's Maven Smart System, which uses AI for image processing and tactical support to suggest and prioritize targets, has reportedly been deployed in recent attacks on Iran. The conflict has also exposed deep ethical divisions over AI's military applications, exemplified by a contract dispute between the US Department of War and Anthropic that culminated in President Trump ordering the government to stop using Anthropic's Claude AI on February 27.
The controversy centers on the US Department of War's January memo requiring AI contractors to allow "any lawful use" of their systems without constraints. Anthropic refused to remove safeguards preventing Claude from being used for mass domestic surveillance or to guide fully autonomous weapons, leading to the contract termination. The company has since been replaced by OpenAI, which agreed to similar restrictions despite the government's broader directive. Political scientist Michael Horowitz notes that rapid technological development is outpacing international regulatory efforts, with academics and legal experts meeting in Geneva this week to discuss autonomous weapons systems.
While proponents argue AI precision targeting could reduce civilian casualties, researcher Craig Jones warns there's no evidence supporting this claim, pointing to high civilian death tolls in Ukraine and Gaza where AI-assisted targeting is used. Current LLM-powered autonomous weapons are not considered reliable enough to comply with international humanitarian laws requiring distinction between military and civilian targets. The conflict has amplified concerns about the proliferation of AI warfare technology before international agreements on ethical and legal uses can be established.
- Experts warn that AI warfare technology is advancing faster than international regulations, with no evidence that AI reduces civilian casualties in conflicts like Ukraine and Gaza
- Current LLM-powered autonomous weapons are not reliable enough to comply with international humanitarian laws requiring distinction between combatants and civilians
