You’ve probably felt that creeping unease. You ask ChatGPT to help brainstorm a half-formed idea, you feed it your best theories, and months later, a shiny new AI breakthrough drops that looks suspiciously like your brainchild. You feel crazy. You feel gaslit. Because you are.
Look at what just happened with OpenAI. A mathematician claims the company essentially stole a major mathematical proof. The internet is doing what the internet does—arguing over the evidence. Some say the claim is weak, just a researcher who had a discussion with an AI about a topic they were working on. But others point out the darker truth: this mathematician explicitly opted out of training data on June 29. They asked OpenAI if they trained on their data. OpenAI said it “did not happen.” Except it clearly did.
When AI companies deny using your data while their models parrot your exact phrasing, they aren’t just stealing your work—they’re gaslighting you into silence.
But let’s step back from the forensic analysis of who said what to a chatbot. The internet is busy arguing about credit, theft, and who gets the Nobel Prize for the next big equation. That’s a distraction. The real danger isn’t that OpenAI might steal your proof. The real danger is what happens next.
If you’re a researcher, an academic, or a creator, you know the stakes. You want to use AI to accelerate your work, but you also know that any early-stage idea you type into that prompt box can be absorbed into the void, repackaged, and spat out as the machine’s own “genius.” We celebrated human-AI collaboration as the ultimate discovery accelerator. Instead, we’ve built a system that punishes transparency.
We are already seeing the perverse incentives this creates. People are now trying to discuss every possible idea with an LLM just to later claim the AI stole it. Meanwhile, tech CEOs use these supposedly “autonomous” breakthroughs to fuel idiotic statements about AGI being achieved, stoking an already dangerous financial fire. It’s a house of cards built on stolen foundations.
The ultimate tragedy of AI isn’t that machines will outsmart us—it’s that humans will stop sharing their brilliance out of fear of being robbed.
This is the behavioral twist nobody is talking about. The true cost of AI opacity isn’t a single disputed mathematical proof. It’s the chilling effect. Why would a researcher openly engage with AI tools on the bleeding edge of discovery if the reward is having their work absorbed without acknowledgment? They won’t. They’ll stop feeding the machine. They’ll hoard their best ideas.
The discovery slowdown won’t be because AI hit a technical wall. It will be because the humans got smart and stopped sharing.
You can’t build the future of human knowledge on a foundation of theft and gaslighting. Eventually, the humans just stop building.
FAQ
Q: Isn't this just a weak claim from a researcher who didn't even have a full proof yet?
A: Whether the proof was finished or not misses the point. The researcher opted out of training, was explicitly told their data wasn't used, and the model clearly used it anyway. The issue is the breach of consent and the gaslighting, not just the final equation.
Q: What's the practical implication for me?
A: If you are feeding proprietary, early-stage, or highly valuable ideas into AI models, you are operating in a gray zone. Assume your prompts and data can be absorbed without credit. Protect your core IP and stop treating LLMs like trusted confidants.
Q: What's the contrarian take?
A: If we put aside credit and ego, human-AI collaboration is genuinely supercharging discovery. The obsession over who gets the applause is a relic of an outdated academic system. The ideas are spreading faster than ever, and the progress itself is what matters.