Pentagon Slashes Key AI and Weapons Testing Team Amid Military Tech Push

Major Cuts to the Operational Test and Evaluation Office

On May 28, Secretary of Defense Pete Hegseth announced significant reductions to the Department of Defense’s Office of the Director of Operational Test and Evaluation (DOTE). This office, responsible for independently testing and evaluating the safety and effectiveness of weapons and AI systems, will see its staff halved from 94 to about 45. The director of the office has been fired and replaced, with only seven days given to implement these sweeping changes.

Background and Importance of the Office

Established in the 1980s following Congressional directives, the DOTE office acts as the final checkpoint before new military technologies are fielded at scale. Former US Navy fighter pilot and engineering professor Missy Cummings emphasizes that this office provides an independent, bipartisan review to reduce waste, fraud, and abuse by verifying contractors’ claims and conducting rigorous safety testing.

The Impact on AI and Military Technology Adoption

These cuts come at a critical moment as the Pentagon rapidly integrates AI into its systems. Companies like OpenAI have become more open to military collaborations, and defense contractors such as Anduril have secured massive funding rounds and contracts to deploy AI technologies. Hegseth argues that streamlining testing will save $300 million and improve efficiency.

Concerns Over Safety and Effectiveness

Experts worry that these reductions may compromise the thoroughness of safety evaluations. Mark Cancian, a senior advisor at the Center for Strategic and International Studies, warned that without comprehensive testing, some problems might only emerge during combat. Missy Cummings points out that the cuts signal an intent to remove obstacles for favored companies, potentially at the expense of safety and efficacy.

Uncertain Outcomes Amid Rapid AI Development

While the military has experimented with AI technologies like computer vision for drones for years, the current focus on large language models and other cutting-edge AI introduces new challenges. These systems are prone to hallucinations and errors, making independent evaluation more crucial than ever. The halving of DOTE’s staff raises questions about the Pentagon’s commitment to thorough testing as it pushes AI deployment forward.

Industry Responses

Neither Anduril nor Anthropic responded to inquiries about their stance on the cuts. OpenAI stated it was not involved in lobbying for the restructuring. The debate continues over how these changes will affect the future safety and reliability of military AI systems.