Newly unsealed material in a major copyright lawsuit alleges that employees and executives at OpenAI and Microsoft recognized that generative AI could pose an “existential threat” to publishers even as the companies continued developing systems trained on enormous quantities of news content. Internal communications cited in the litigation describe AI products as increasingly substituting for publishers and warn of a “doom loop” in which AI reduces traffic and revenue to the very organizations producing the information AI systems depend upon. The filings also allege large-scale copying of copyrighted and paywalled material, while Microsoft’s own data reportedly showed steep reductions in click-through traffic to some publisher sites when AI-generated answers replaced conventional search. OpenAI and Microsoft maintain that model training can constitute fair use, making the dispute a potentially defining test of how traditional copyright and property rights apply when AI businesses derive commercial value from material created by others.
Key Takeaways
- Internal communications cited in the litigation indicate that personnel inside OpenAI and Microsoft understood that AI systems could become substitutes for news publishers, potentially damaging the businesses whose reporting supplied critical material for AI development.
- Microsoft data cited in the filing reportedly found click-through declines as high as 93% for certain publisher domains when users encountered its AI answer engine instead of conventional search; independent research has similarly found that users presented with AI summaries click traditional search results considerably less often.
- The dispute could have consequences far beyond the parties involved because it addresses a fundamental question for the AI economy: whether mass use of copyrighted material to train commercial AI systems constitutes permissible fair use or infringement requiring authorization and compensation.
In-Depth
The newly unsealed material raises a question more consequential than whether artificial intelligence is innovative: whether companies can build valuable products by consuming publishers’ work while weakening the economic system that produces it. Internal communications cited in the litigation indicate personnel at OpenAI and Microsoft understood that risk. Microsoft researcher Brent Hecht reportedly warned of a “doom loop” in which AI products damage their content supply chain, while OpenAI executive Nick Turley described increasingly capable chatbots as substitutive for publishers.
That concern is no longer theoretical. Microsoft data cited in the filing reportedly showed dramatic declines in click-through rates to certain publisher websites when users received AI-generated answers. Separate research on AI search summaries likewise found lower click-through rates when generated answers appear above conventional results. The economic problem is straightforward: reporting costs money, while automated systems can summarize reporting at scale without bearing the original expense of producing it.
The legal issue remains unsettled. OpenAI and Microsoft maintain that training models on copyrighted material can qualify as fair use, while publishers argue unauthorized copying becomes especially problematic when resulting products compete directly with businesses supplying the training material.
The conservative principle at stake should be familiar: property rights do not become obsolete because technology makes appropriation efficient. Innovation deserves room to flourish, but functioning markets also require enforceable ownership, voluntary exchange and meaningful compensation. If AI companies need journalism to create valuable products, licensing offers a market-based mechanism for preserving technological progress and the institutions that generate original information.
Sources
- https://techcrunch.com/2026/09/17/microsoft-exec-called-ai-scraping-the-largest-theft-of-labor-in-human-history-new-unredacted-filings-reveal/
- https://arstechnica.com/tech-policy/2026/09/microsoft-exec-called-ai-scraping-the-largest-theft-of-labor-in-human-history/
- https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
- https://www.reuters.com/legal/litigation/us-judge-says-she-will-stay-openai-case-after-divesting-microsoft-stock-2026-09-14/