AI execs admit it: the biggest theft on the web‘Doom Loop’: OpenAI and Microsoft Admits LLMs Are Destroying the Web and Built on Theft

The New York Times has been suing OpenAI over copyright for years. Last Thursday a filing came out unredacted, and it is mostly Microsoft and OpenAI executives talking about their own products. Both companies had asked for those passages to be sealed or blacked out.

How big the theft was

The Times' lawyers dug out an internal Microsoft document that puts it plainly: millions of people around the world will soon see large models hoovering up their work as an astonishing theft of unprecedented proportions.

It adds that almost no one meant their writing to be used this way, and nobody was paid. One Microsoft executive called it the largest theft of labour in human history.

Steal from them, then starve them

The second half is worse. An internal Microsoft document admits the company's AI content strategy started a doom loop.

It grabs writers' work and it grabs their clicks. The phrasing is precise: it is highly unusual for an end product to threaten the economic foundations of its own suppliers.

Nadella and others testified under oath that once the content was taken, clicks through to news sites on Bing fell by more than 90 percent.

One OpenAI document records a hack to get around the New York Times paywall. Co-founder Greg Brockman replied: ah, nice.

OpenAI's policy chief Jack Clark wrote that the company was building systems that substitute for the labour of the people who define a society's culture.

OpenAI called itself an existential threat to news publishers. One engineer testified that no matter how prominently the links are shown, users will not click.

In public, both companies' lawyers argue in court that the training is fair use, transformative. In the emails, it reads differently.

Strip the jargon and it is one sentence: they took your work, killed your business, and hired lawyers to say it is not theft.