Microsoft Employees Debate AI Scraping as Major Labor Theft

Summary

• Internal Microsoft memos raise concerns about AI scraping as potential labor theft.
• CEO Satya Nadella asserts that paywalled content should be licensed for use.
• Documents are part of a lawsuit involving the New York Times against OpenAI and Microsoft.

Internal discussions at Microsoft have sparked a debate among employees regarding the ethics of OpenAI's use of news articles, with some labeling it as possibly the "largest theft of labor in human history." This revelation comes from court documents released in connection with a copyright lawsuit filed by the New York Times against OpenAI and Microsoft.

The lawsuit, initiated in late 2023, has expanded to include eleven additional publishing entities. OpenAI has consistently denied the allegations and is required to maintain 20 million ChatGPT conversation logs as part of the ongoing legal proceedings. Judge Sidney Stein from the Southern District of New York is currently reviewing motions for summary judgment as more documents are disclosed.

A memo from Microsoft, authored by Brent Hecht, an applied science director, suggested that the public might view AI models as "hoovering up" their work, constituting an unprecedented form of theft. Microsoft has clarified that Hecht's views do not represent the company's official position, as he was tasked with providing varied perspectives rather than making decisions.

Licensing Content

During testimony, Satya Nadella emphasized that any content behind a paywall should be licensed by users, stating that had he known OpenAI was utilizing such material, he would have advocated for retraining the models. This reflects a broader principle regarding information access and consumption.

Meanwhile, an OpenAI employee informed president Greg Brockman about a method to bypass the New York Times paywall, which Brockman acknowledged positively. Concerns about AI's impact on the publishing industry have been echoed by figures like Nick Turley, who noted that AI could pose an "existential threat" to traditional publishers as it evolves.

Additionally, a former OpenAI policy director, Jack Clark, warned in a 2020 memo about the potential for AI systems to replace human labor in cultural contributions, highlighting the risks of unchecked technological advancement. Clark later co-founded Anthropic, which did not comment on the current situation.

Both companies assert their training practices are compliant with fair use, arguing that they transform existing articles into new works. Steven Lieberman, representing the New York Daily News and other publications involved in the lawsuit, remarked on the implications of OpenAI's and Microsoft's views regarding their practices.

The New York Times, a key plaintiff in the case, has refrained from commenting, while OpenAI has not responded to requests for further information.