AI-generated editorial illustration; not a documentary photograph.
A January 2023 memo by Microsoft Director of Applied Science Brent Hecht, disclosed in recently unredacted court filings, described AI scraping as a large-scale appropriation of human labor. TechRadar reports that the memo surfaced in The New York Times' copyright case against Microsoft and OpenAI. It is an internal warning, not a court finding about whether the companies' use of published work was unlawful.
The Internal Warning
Hecht argued that generative AI could disrupt the jobs of people whose work had been collected for training without their consent. His concern connects two issues for publishers and creators: the use of their material to build AI systems and the possibility that those systems could reduce demand for the original work.
Publisher Traffic
TechRadar also reports that Microsoft's internal research found Copilot could reduce click-throughs to The New York Times by as much as 93% compared with Bing's conventional search results. That is a reported potential reduction in a specific comparison, not a measured decline across the publisher's traffic. It illustrates why publishers may be concerned about AI answers even apart from how models are trained.
Claims in the Case
The New York Times' case accuses Microsoft and OpenAI of copying and using copyrighted work without permission. According to TechRadar, the case also alleges that OpenAI's mid-training datasets contained more than 91,000 works from The New York Times, Daily News and the Center for Investigative Reporting. A separate allegation says OpenAI removed copyright notices from material before it entered large datasets. These are claims in litigation, not established findings.
What Remains Open
TechRadar reports that Microsoft CEO Satya Nadella said he would have exercised Microsoft's rights to require OpenAI to retrain models without paywalled data had he known OpenAI was collecting it. That hypothetical does not establish that retraining occurred. The case against OpenAI remains open, and the legal treatment of copyrighted material used for AI training remains contested.
SIGNAL TITANS
Leave a Reply