AI Firms Forge Closer Ties with US Military, Intelligence Amid Policy Shifts
TEHRAN (Tasnim) - Anthropic and Meta have adjusted their policies to facilitate military access to their artificial intelligence technologies, underscoring a shift among tech giants to support defense applications despite some internal opposition.
Leading AI companies, traditionally hesitant to engage in military partnerships, are increasingly accommodating requests from the Pentagon and intelligence agencies.
On Thursday, AI start-up Anthropic announced a deal to provide its technology to US military and intelligence agencies through Amazon’s cloud and government software provider Palantir, according to WashingtonPost.
This shift follows Meta's policy update on Monday, enabling military applications of its open-source AI model, Llama. The same day, OpenAI signed a deal to supply ChatGPT to the US Air Force, having previously adjusted its stance to allow certain military uses.
These developments highlight the tech industry’s growing alignment with defense agencies, though some employees have voiced ethical concerns.
In June, Anthropic modified its policies to allow intelligence uses while maintaining restrictions on weaponry and domestic surveillance applications. OpenAI similarly bars its technology from being weaponized. Both firms referred only to their policies when asked for comments.
Arms-control advocates continue to call for a global ban on AI weapons, though the US military insists on human oversight for AI-driven weapon systems but has refrained from an outright ban.
Amid rising tensions with China, US tech executives and lawmakers are emphasizing the need to advance AI for defense to ensure national competitiveness. Anthropic CEO Dario Amodei argued in a recent blog post that democratic nations should develop leading AI capabilities to counter autocracies that could use such technology for human rights abuses.
Anthropic has received backing from tech giants Google and Amazon, with a $4 billion investment. Amazon founder Jeff Bezos owns The Washington Post.
AI is currently used in the military for diverse applications, such as predicting aircraft maintenance needs and identifying potential targets.
Partnering with Palantir, Anthropic aims to deliver its technology to government clients. Palantir's AI tools, used to detect targets from satellite and aerial imagery, remain a cornerstone of its offerings. The firm’s adviser, Jacob Helberg, attended the recent election event for Republican candidates.
The conflict in Ukraine has sparked interest in adapting commercial technology, such as small drones and satellite internet, for military purposes. Numerous Silicon Valley start-ups have emerged, seeking to modernize defense offerings with innovative tools.
Military leaders worldwide predict AI will reduce reliance on human oversight in future combat technology. Despite ongoing human control over critical decisions, arms-control advocates caution against potential errors and violations of international law.
Competition among Google, Microsoft, and Amazon for military cloud-computing contracts is intense, though employee opposition to defense work persists. In 2018, Google chose not to renew a Pentagon contract following employee protests, though the company has since continued expanding its defense contracts.
This year, Amazon and Google faced employee objections over contracts with the Israeli government, with some Google DeepMind staff urging the company to avoid selling AI to militaries.
Companies like OpenAI and Anthropic are embracing defense work earlier in their lifecycle, while other AI-focused firms, such as Scale AI, have prioritized military partnerships as core to their business.
