The Memo Heard Round the Boardroom
Even now, somewhere in a Redmond conference room with excellent snacks, someone at Microsoft is explaining that words have meanings, except when they don't, depending on who said them and to whom. That's the shape of the defense forming around a January 2023 internal memo in which Brent Hecht, the company's Director of Applied Science, described its approach to training data as "an astonishing theft of unprecedented proportions" — and, for those who prefer their appraisals maximalist, "the largest theft of labor in human history."
One presumes Hecht did not intend for this document to surface during discovery in a copyright suit filed by The New York Times. And yet — as if this were not enough — it did, unsealed on September 17 as part of the Times's summary-judgment brief against Microsoft and OpenAI.
Sources confirm the memo was written in 2023. The lawsuit was filed the same year. The math on how long a company can sit on its own executive's candor, dear reader, checks out.
The Algorithm's Other Half Weighs In
Not to be left out of the self-incrimination, OpenAI contributed its own internal assessment, courtesy of Nick Turley, Head of ChatGPT, who wrote that the company's products are "largely substitutive, period" — legal-adjacent phrasing for "yes, we built the thing that replaces the thing." Turley reportedly went on to warn that chatbots pose an "existential threat" to publishers, which is one way of putting what publishers have said for three years without an internal memo to back it up.
Deposition testimony from Microsoft CEO Satya Nadella apparently agrees: chatbot conversations, he conceded, had "substituted" for visits to publisher websites. The Times alleges Bing Copilot alone dropped its click-through rate by 83 to 93 percent, alongside claims of paywall bypassing and stripped copyright notices — the sort of detail that reads less like a legal brief and more like an inventory.
In a development that will surprise no one who has been paying attention, the people building the theft-machine kept excellent internal records of it being a theft-machine.
The Fair-Use Community Responds
Asked to reconcile "largest theft of labor in human history" with its official legal position, a Microsoft spokesman offered that Hecht's comments reflected "one employee's individual perspective" — as though the company's Director of Applied Science moonlights as a freelance op-ed columnist with no bearing on what the company actually did with the data. OpenAI, given the same opportunity, declined to comment at all, which is its own kind of answer.
The case — The New York Times v. Microsoft and OpenAI, filed in the Southern District of New York before Judge Sidney Stein — now awaits a summary judgment ruling that could decide which claims survive to trial. The Fair Use Community, ever resourceful, remains confident that calling something theft in writing is not the same as it being theft, provided enough lawyers agree.
One presumes discovery has more of these. It usually does.
Sources: TechCrunch — Microsoft exec called AI scraping 'the largest theft of labor in human history,' new unredacted filings reveal · MLQ News — Unsealed New York Times brief exposes internal AI-substitution warnings from Microsoft and OpenAI
