OpenAI Robotics Chief Quits Over Military AI Ethics, Sparking Internal Rift

MarketDash

Mar 7, 8:56 PM

Caitlin Kalinowski's resignation from OpenAI highlights a growing tension within AI labs over military contracts, surveillance, and autonomous weapons—just as the company expands into classified Pentagon work.

Key Points

Caitlin Kalinowski, head of OpenAI's robotics division, resigned over concerns that AI uses for domestic surveillance and autonomous weapons weren't debated thoroughly enough.
Her departure comes as OpenAI secures a Pentagon deal that includes guardrails against mass domestic surveillance and requires human control over force.
The move contrasts with rival Anthropic, which was blacklisted by the Trump administration for refusing to drop similar safety clauses from its own agreement.
OpenAI CEO Sam Altman says the company's shift to classified work is urgent and complex, and that the Pentagon accepted its core principles as consistent with U.S. law.
The situation reveals a split in how AI companies navigate national security contracts, balancing ethical boundaries with government demands.

Get Market Alerts

Weekly insights + SMS alerts

Here's a story about what happens when the people building the robots start worrying about what the robots might do. Over the weekend, Caitlin Kalinowski—the leader of OpenAI's robotics division—said she quit. Her reason? She thinks the company didn't spend enough time talking about whether AI should be used to spy on Americans without a warrant, or to let weapons make lethal decisions without a human in the loop.

That's a pretty specific set of concerns. And it's especially interesting because it comes just as OpenAI is getting deeper into classified work with the Pentagon. The company has a new arrangement that, according to CEO Sam Altman, includes two big limits: no domestic mass surveillance, and a requirement that humans stay in control of any use of force. So on paper, the guardrails Kalinowski is worried about are already there. But for her, it seems the conversation around them wasn't.

In a post on X, she put it plainly: "I resigned from OpenAI. I care deeply about the Robotics team and the work we built together. This wasn't an easy call. AI has an important role in national security. But surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got. This was about principle, not people. I have deep respect for Sam and the team, and I'm proud of what we built together."

So it's not that she's against AI in national security—she says it can matter. It's that she wants those two lines drawn in the sand to be debated more carefully. And she's framing this as a values thing, not a personal beef with Altman or the team.

Why This Resignation Matters

Kalinowski's exit isn't just one person's career move. It touches on the exact fault lines that are shaping how top AI labs deal with the U.S. national security world: surveillance at home, and autonomy in weapons. She's saying those issues didn't get the weight she expected.

Meanwhile, Altman has been talking about how OpenAI's stance has shifted. They used to avoid classified gigs; now they're taking them on with the Department of War. He calls the shift urgent and more complex than earlier work. He also mentioned that OpenAI previously passed on classified opportunities that rival lab Anthropic accepted.

Under the Pentagon deal, OpenAI kept those two guardrails—no domestic mass spying, human control over force—while adding operational stuff like putting engineers on-site to watch how models behave. Altman says the company will build technical constraints to keep systems within expected limits, and that the Department of War wants those protections too.

The Anthropic Angle: A Tale of Two AI Labs

Here's where it gets even more interesting. Kalinowski's resignation comes right after OpenAI's Pentagon deal, which was signed just hours after the Trump administration blacklisted Anthropic for refusing to take similar safety clauses out of its own agreement. So one company gets a deal with its principles intact; the other gets labeled a supply chain risk.

When Anthropic wouldn't budge, Defense Secretary Pete Hegseth called the company a supply chain risk, and President Donald Trump told agencies and military contractors to cut ties. Anthropic said it was "deeply saddened," called the designation "legally unsound," and warned it would "set a dangerous precedent for any American company that negotiates with the government."

Altman says OpenAI negotiated so that other AI developers could get comparable terms, not just his firm. But the outcome is still stark: OpenAI says it got the Pentagon to accept its two red lines, while Anthropic got blacklisted even though it described similar boundaries.

Altman frames this as the Department of War seeing OpenAI's principles as consistent with existing U.S. law and policy. He says the quick move was about avoiding what he sees as a dangerous competitive spiral among AI labs. Kalinowski's resignation, on the other hand, shows how internal talent might react when those same boundaries feel like they weren't examined enough.

So you've got two stories here. One is about a company navigating government contracts and trying to set a precedent. The other is about an engineer who built robots deciding that the ethical conversation around where those robots might go wasn't loud or long enough. And both are happening at the same time, in the same industry, as AI starts to move into places where the stakes get very, very real.