Copyright Infringement Still Isn’t Theft, Even When A Microsoft Employee Says It Is

The article argues that recent media frenzy surrounding an Microsoft employee’s characterization of AI data scraping as “theft” is legally misguided and emotionally driven rather than factually grounded. It emphasizes that copyright infringement is distinct from theft, as infringement involves copying without license rather than removing physical possession. Furthermore, it highlights the strong legal argument that AI training constitutes fair use, noting that such transformative uses are not copyright infringement at all, regardless of how industry players or employees colloquially frame the issue. The piece critiques the New York Times’ lawsuit strategy, suggesting it relies on cherry-picked, out-of-context quotes from corporate filings to manipulate public perception and legal narratives. It asserts that competitive market pressures or user preference for AI summaries over original articles do not constitute illegality. The author contends that equating economic competition with copyright violation ignores the fundamental purpose of fair use, which is designed to balance rights with public interest and innovation, rather than protecting established business models from disruption. This discussion is critical for open data because it challenges the narrative that machine learning training on publicly available data is inherently unlawful or unethical. By clarifying that fair use is a right, not a defense, and that “theft” is a misleading term, the article supports the legitimacy of using open sources to train AI models. It encourages the open data community to focus on the nuanced legal realities of fair use rather than accepting inflammatory rhetoric that seeks to restrict access to and use of public information for technological advancement.

Source: techdirt.com
Published on 2026-09-24