OpenAI and Microsoft knew they were starting a ‘doom loop’ for the web

The courthouse is becoming the only place where the AI industry’s internal contradictions get an honest airing. The recently unsealed documents in the <i>Times</i> v. OpenAI case are striking not because they reveal an unknown scandal, but because they show how clearly the companies understood the game they were playing. They didn’t stumble into a broken internet — they logged the warning signs, coined a phrase for the eventual collapse (“doom loop”), and proceeded anyway. “Doom loop” deserves genuine attention as a technical and economic concept. It describes a feedback spiral: generative AI floods the web with synthetic content, human-produced material gets buried and devalued, publishers stop producing it, and future AI models have no fresh human data to train on. The result is a web that is simultaneously bloated and hollow. OpenAI’s own materials apparently predicted this dynamic with clinical precision — which makes the public dismissal of existential AI risks feel less like naivety and more like compartmentalization. Then there is the phrase “largest theft of labor in human history.” It is hard to imagine a more damning description of the foundational act of commercial AI development. The language is hyperbolic, sure, but it is also precise: the value of large language models derives overwhelmingly from human intellectual labor that was appropriated without consent or compensation. The fact that those words ended up in internal memos, rather than in marketing copy, reveals a workforce and leadership that knew the moral ledger was deeply in the red. The unsealed filings don’t resolve the legal questions, but they do reframe them. This was never about a few copyrighted articles slipping into a training set. It was about a systematic industrial strategy with known consequences for writers, publishers, and the information ecosystem at large. The <i>Times</i> case, detailed by <i>The Verge</i> at the source link below, is among the first to force the industry to explain those choices under oath. And when the curtain is pulled back, the explanation seems to be: we saw the cliff, and we accelerated anyway. Source: [https://www.theverge.com/ai-artificial-intelligence/997633/openai-microsoft-chatgpt-ai-new-york-times-doom-loop-theft-google-zero](https://www.theverge.com/ai-artificial-intelligence/997633/openai-microsoft-chatgpt-ai-new-york-times-doom-loop-theft-google-zero)