Elon Musk testifies that xAI trained Grok on OpenAI models

So here’s something interesting happening behind the scenes in the AI world right now. Big AI companies spend hundreds of millions of dollars training their models on massive amounts of data. But there’s a shortcut some smaller players use called “distillation” — basically, you let a big expensive model generate a bunch of outputs, then train your cheaper model on those outputs instead of raw data. Think of it like a student who learns not from textbooks but by copying a really smart classmate’s homework. You end up with a capable student without doing all the original reading yourself. Elon Musk essentially admitted his company did something like this with Grok, using OpenAI’s models as part of the training process — which is awkward given his very public beef with OpenAI.

Why does this matter to you? Because it’s kicking off a real fight over who owns AI “knowledge.” The big labs are now writing terms of service that explicitly ban using their outputs to train competing models, and they’re getting better at detecting when it happens. It’s a bit like a restaurant trying to legally stop you from reverse-engineering their secret sauce. This legal and technical tug-of-war will shape which AI tools survive, which get sued into oblivion, and ultimately which ones you’ll have access to in two years.

Here’s where you can actually use this to your advantage. First, if you’re a small business owner, pay attention to which AI tools are building on solid legal ground versus which ones might disappear or get hobbled by lawsuits. Tools built on distilled or legally shaky models could vanish or degrade overnight, so diversify what you rely on rather than going all-in on one platform. Second, there’s genuine freelance work emerging here — companies need people to audit and document how they’re using AI outputs, making sure they stay within terms of service. If you’re detail-oriented, offering “AI compliance” checks to small businesses is a real niche with real demand right now. Third, if you create content for a living — writing, coaching, courses — this is your reminder that human-generated, experience-based content is becoming more valuable, not less, as AI-generated stuff floods the market. Double down on your personal stories and specific expertise, because that’s genuinely hard to copy or distill.

The messy legal reality of AI training means the smartest move is treating these tools like borrowed equipment — useful today, but always have a backup plan.