Caitlin Kalinowski resigned from OpenAI on March 7, 2026. She wasn’t a mid-level manager. She was the executive leading OpenAI’s entire robotics division — the person Sam Altman trusted to build the company’s physical AI future. Her reason was direct: the Pentagon deal announced just over a week ago moved too fast, with too few guardrails on surveillance and autonomous weapons.

“Surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got,” she wrote publicly on LinkedIn. OpenAI confirmed the departure. No pushback on her account of events. Just a statement about “workable paths” and “red lines.”

The media is covering this as an ethics story. It isn’t — or at least, not primarily. This is a trust collapse with direct financial consequences for anyone whose business or career depends on AI tools.

Here’s the number that matters: ChatGPT uninstalls surged 295% in the days after the Pentagon deal was announced. Claude simultaneously climbed to the #1 spot in the U.S. App Store. That’s not a blip — that’s users making an active, emotional decision about which company they trust with their data. When a senior executive publicly validates those fears by walking out the door, that signal gets louder. Businesses that built workflows around ChatGPT now face a credibility question from their own clients: “Are you sure you want our data running through that?”

The deeper issue is governance speed. Kalinowski’s sharpest critique wasn’t the deal itself — it was that the announcement was rushed without defined guardrails. That pattern, moving fast and defining the rules later, is exactly how enterprise procurement officers justify switching vendors. And switching costs are real: teams retrain, integrations break, productivity dips for weeks.

If your business or employer uses ChatGPT at scale, this week is the right moment to audit your AI vendor exposure — not out of panic, but out of professional diligence.

One concrete action: Pull up whatever AI tools your team uses and check whether your data is being used to train models. OpenAI’s enterprise tier offers opt-out; the free and Plus tiers do not by default. If your company hasn’t locked in enterprise data protections, this story gives you the business case to raise it with whoever controls the IT budget. Frame it as risk management, not politics. Because that’s exactly what it is.

The companies that win in the next 18 months won’t necessarily use the most powerful AI. They’ll use the AI their clients trust the most.