Senator Questions Pentagon's Influence Over AI Safety Policies

A U.S. senator is demanding the Pentagon explain whether it is pressuring artificial intelligence companies to relax their safety rules. Senator Edward Markey, a Massachusetts Democrat, sent a formal letter to Defense Secretary Pete Hegseth in June 2025, citing troubling reports of officials intimidating AI vendors. The senator's inquiry focuses on whether companies are being pushed to weaken their acceptable use policies—internal rules that often prohibit AI from being used in autonomous weapons or unchecked surveillance—as a condition for securing government contracts.

The letter arrives as Anthropic, an AI firm once celebrated for its strict safety principles, deepens its defense work. The company, creator of the Claude AI models, has recently revised its policies to allow certain military and intelligence applications and has secured contracts with agencies through partners. This shift has sparked debate within the industry and led to reported staff departures from Anthropic's safety teams.

Markey's broader concern is that the Defense Department's immense purchasing power could systematically dismantle the fragile, voluntary safety standards that currently guide the AI industry. With no comprehensive federal AI law in place, these company policies are the primary guardrails. The Pentagon, meanwhile, is rapidly expanding its use of AI for everything from logistics to intelligence, and officials have publicly suggested that restrictive policies put vendors at a disadvantage.

For AI companies facing steep costs and investor pressure, the defense market's projected billions are a powerful lure. Critics warn this creates a race to the bottom on safety ethics. Markey has given the Pentagon until July 3, 2025, to answer detailed questions about its practices. The response, or lack thereof, will test the credibility of both the industry's self-regulation and the government's procurement ethics.

Source: Webpronews