Hard Fork

A.I. Goes to War + Is ‘A.I. Brain Fry’ Real? + How Grammarly Stole Casey’s Identity

with Julie Bedard
13 Mar 2026 16 min read 1h 2m
TL;DR

AI is now core to military operations—Claude specifically is suggesting targets and streamlining battlefield decisions in the Iran conflict—while simultaneously draining workers who oversee these systems. The shift reveals how tech companies quietly abandoned their anti-military principles, and how the cognitive load of AI oversight is creating a new form of workplace exhaustion researchers call 'AI brain fry.'

Key Moments
Kevin Roose
“tools that are deployed abroad during times of war sometimes come back home after the war and wind up being used against American citizens.”
Explaining why covering AI in military conflict matters beyond geopolitics
▶ 1:15
Casey Newton
“Claude has been integrated into Maven Smart System since 2024, and the reporting that I've seen over the past week, including in this article in the Washington Post, said that this combination of the Maven Smart System built by Palantir and Claude has already suggested hundreds of targets, issued precise location coordinates, and prioritized those targets according to importance.”
Detailing the specific role Claude plays in real-time military operations
▶ 10:25
Kevin Roose
“all of these companies were run by people who at one point thought this was all a bad idea to be selling these very advanced AI tools to the military. And then they changed their minds, and they did that because of some combination of pressure or just maybe market opportunity to get these big military contracts, but they did at one point have a principle that involved we don't want our stuff being used to kill people.”
Reflecting on how OpenAI, Google, Meta, and others reversed their anti-military policies
▶ 19:01
Julie Bedard
“14% of people who use AI said that they felt this. And I was especially surprised by the extent to which they told us about it. We asked, you know, free-ended, like, just tell us, what is this thing? What does it show up? How does it feel to you? Um and people wrote a lot, right? Like, they wrote all these things about, feels like I have 12 browser tabs open in my head.”
Describing survey findings on AI brain fry prevalence and how workers describe the experience
▶ 24:35
Julie Bedard
“A marketing manager was 90% disrupted from a skill perspective. So, so that's sort of the first fundamental piece about marketing is like they've tended to adopt and it's a really different way of working because of the power of the tools.”
Explaining why marketing roles experience the most AI brain fry compared to other industries
▶ 29:07
About the show
›
Hard Fork is the New York Times' podcast about the internet, big tech, and how it shapes our world. Hosts Kevin Roose and Casey Newton dive deep into AI deployment in military conflict, the psychological toll of AI tools on workers, and the corporate principles that shift under pressure.
Takeaways
1
Claude is now operational in classified military systems Anthropic's Claude is the only large language model currently deployed inside U.S. classified military systems and is central to targeting decisions through Palantir's Maven Smart System. The tool can suggest hundreds of targets with precise coordinates in real-time, transforming weeks-long planning into live operations—making it difficult for the military to abandon even amid Anthropic's stated concerns about Pentagon use.
2
Tech companies abandoned anti-military principles quietly DeepMind, Google, OpenAI, and Meta all previously prohibited military applications in their terms; all have now quietly reversed those policies between 2024-2025. This shift happened under pressure and market opportunity rather than principled reassessment, suggesting that company ethics statements may be vulnerable to erosion during competitive or geopolitical pressure.
3
AI oversight creates distinct cognitive exhaustion in workers Research shows 14% of AI users experience 'AI brain fry'—cognitive strain from excessive oversight and iteration beyond their cognitive capacity. This is distinct from burnout and hits hardest in marketing roles (90% skill disruption) where workers must constantly iterate and validate AI outputs without clear success thresholds, leading to information overload and reduced actual work output.