Musk vs. Altman in Court, and the AI Industry’s Underpants Problem

Elon Musk and Sam Altman are finally going to court this week, and the outcome could reshape the entire AI landscape. Musk, an original OpenAI co-founder, claims he was tricked into funding the company under false pretenses. He wants $134 billion in damages, Altman and Greg Brockman out, and OpenAI returned to non-profit status. Ahead of OpenAI’s planned IPO, the court could decide whether the company can even exist as a for-profit enterprise. That’s a lot riding on a single trial.

I’ve been watching this feud simmer for years, and it’s always felt more personal than legal. But the stakes here are real. If Musk wins, it could upend the global AI race—not just OpenAI’s trajectory, but how other AI startups structure themselves. If he loses, it sets a precedent that founders can pivot from non-profit to for-profit without much consequence. Either way, the trial is worth paying attention to.

Meanwhile, the AI industry has a different kind of problem: profit. Will Douglas Heaven at MIT Technology Review nails it with a South Park reference. Remember the underpants gnomes? Phase 1: Collect underpants. Phase 2: ? Phase 3: Profit. That’s basically where AI is right now. Companies have built impressive tech (Step 1) and promised transformation (Step 3), but the middle step is still a giant question mark.

This is higher than I expected, honestly. We’ve seen massive investment in AI infrastructure, but the business models are shaky. Subscription tiers, API pricing, enterprise deals—none of it has proven sustainable yet. The hype is real, but the profit isn’t. And that’s a problem when investors start asking for returns.

Then there’s the deepfake problem. For years, experts warned this was coming. Now it’s here. Cheap, accessible models are producing weaponized deepfakes—from sexually explicit images to political propaganda—that look startlingly real. They’re already inciting violence, changing minds, and sowing mistrust. Women and marginalized groups are disproportionately affected.

What’s alarming is how fast this escalated. A few years ago, deepfakes were obvious fakes with weird blinking and mismatched skin tones. Now they’re good enough to fool most people. And the tools are getting cheaper and more accessible by the month. Trust is eroding, and critical thinking is taking a hit. I’m not sure regulation can catch up fast enough.

In other news, OpenAI ended its exclusive partnership with Microsoft, allowing it to court rivals like Amazon. Microsoft will still license OpenAI’s tech, but no longer exclusively. This is a big shift—OpenAI needs more revenue streams ahead of its IPO, and locking itself into one cloud provider was limiting. But it also signals that OpenAI is missing key growth targets, which is worrying for a company valued at hundreds of billions.

Google signed a classified AI deal with the Pentagon, permitting AI use for “any lawful government purpose.” Over 600 Google workers called for a block on the deal, but it went through anyway. AI firms are now training military versions of their models on classified data. This is a slippery slope, and I’m not comfortable with it. The line between defense and offensive AI is blurring fast.

The original article also mentioned a few other stories I won’t go into detail on, but the pattern is clear: AI is moving from hype to reality, and reality is messy. Legal battles, profit gaps, deepfake crises, military deals—this is the messy adolescence of a technology that promised to change everything. It is changing things, just not always in the ways we expected.