I don't usually talk about politics. It's not what this blog is about, and frankly, it's not where I'm most comfortable writing. But as someone who works in the AI space, I feel a responsibility to lay out what's happening right now, because the intersection of artificial intelligence and politics has reached a point that affects all of us, whether we build AI, use AI, or simply live in a world shaped by it.

Here's what's been going on.

The Anthropic-Pentagon standoff

One of the most consequential stories in AI right now is the public fallout between Anthropic, the maker of the Claude AI model, and the United States government.

Since 2024, Anthropic had a contract worth up to $200 million with the U.S. Department of Defense (which was given the secondary title "Department of War" by executive order in September 2025). Claude was deployed on the Pentagon's classified networks as part of the Maven Smart System, which supports intelligence gathering, image processing, and tactical operations.

In January 2026, the Pentagon issued a memo requiring that all AI procurement contracts allow models to be used "for any lawful purpose," without restrictions imposed by the AI vendor. Anthropic pushed back on two specific points. The company insisted on contractual safeguards that would explicitly prevent Claude from being used for:

Anthropic's CEO, Dario Amodei, argued that current frontier AI models are not reliable enough to be trusted in fully autonomous weapons, and that mass domestic surveillance of Americans represents a fundamental rights violation. The Pentagon's position was that existing U.S. law already prohibits these uses, making Anthropic's additional restrictions unnecessary and an overreach by a private company.

The timeline of escalation

Here's how events unfolded:

OpenAI's Pentagon deal

Hours after the Anthropic ban, OpenAI published a detailed breakdown of its agreement with the Department of Defense for classified systems. The company outlined three "red lines" that it said guide its government work:

OpenAI's deal is cloud-only, meaning the company retains control of its safety stack and does not deploy models on edge devices (which would be required for powering autonomous weapons). Cleared OpenAI engineers and safety researchers remain in the loop.

Notably, OpenAI CEO Sam Altman publicly stated that his company shares Anthropic's red lines. On March 2, OpenAI amended its deal to add explicit language stating that its tools "shall not be intentionally used for domestic surveillance of U.S. persons and nationals." OpenAI also stated publicly that it does not believe Anthropic should be designated a supply chain risk.

The situation raises an uncomfortable question: if both companies share the same red lines, why did one get a deal and the other get banned?

AI lobbying is now big money

Behind the scenes, both companies have dramatically increased their spending on government lobbying.

As Forbes reported in February 2026, "AI's biggest builders are now its biggest lobbyists."

AI in the Iran conflict

On February 28, 2026, the United States and Israel launched a military offensive against Iran. This conflict has become the largest real-world test of AI-assisted warfare to date.

According to U.S. Central Command, AI technology has played a critical role in the campaign:

As Nature reported on March 5, the conflict "has thrown a spotlight on the use of artificial intelligence in warfare." The same week, academics and legal experts met in Geneva to discuss lethal autonomous weapons and AI procurement in the military, as part of ongoing international efforts to establish legal frameworks for AI in warfare.

The possibility that AI's precision targeting could help reduce civilian casualties is often cited as a potential benefit. However, experts note there is currently no evidence that AI lowers civilian deaths, and in recent conflicts where AI has been used to assist targeting, civilian death tolls have remained high. "There is no evidence that AI lowers civilian deaths or wrongful targeting decisions and it may be that the opposite is true," said Craig Jones, a political geographer at Newcastle University.

The deepfake crisis

Perhaps the most immediately dangerous intersection of AI and geopolitics right now is the flood of AI-generated disinformation tied to the Iran conflict.

This is real, it is happening at scale, and it is affecting what people believe about the war:

The scale of the problem is staggering. Industry estimates suggest the number of deepfake videos shared online surged from approximately 500,000 in 2023 to 8 million by 2025. Research indicates that 68% of deepfakes are now "nearly indistinguishable from genuine media." And studies have shown that even when people are explicitly warned that a video is a deepfake, most still rely on the content of the video to form their beliefs.

Please verify your sources. If you see dramatic footage of the conflict, check who posted it, when it was posted, and whether credible news organizations have confirmed it. Use tools like reverse image search, check fact-checking organizations (like BBC Verify, which has been actively debunking AI-generated content from this conflict), and be especially skeptical of content that seems designed to provoke an emotional reaction.

The market impact

The convergence of AI uncertainty and geopolitical conflict has rattled financial markets:

Separately, AI-specific fears have also weighed on stocks. In late February, a hypothetical "macro memo from 2028" about how AI could disrupt the economy sent the Dow down 800 points, illustrating how sensitive markets have become to AI-related narratives.

What this means

Here are the facts as they stand:

None of this is theoretical anymore. These are things happening right now, in real time, affecting real people and real decisions.

I wrote this post because I believe that anyone working in or around AI has a responsibility to understand the broader context in which this technology exists. I don't have easy answers, and I'm not here to tell anyone what to think. But I do believe that being informed, being skeptical, and being honest about what's happening is the bare minimum.

Stay critical. Verify everything. And if someone shows you a dramatic video from a war zone, ask yourself: is this real?

References