Publisher group cites Microsoft’s ‘doom loop’ in OpenAI case

Digital Content Next (DCN), a trade group whose members include The New York Times, asked a federal judge in New York on October 29, 2026 to accept an amicus brief in the consolidated OpenAI copyright litigation. The brief pairs Microsoft’s own “doom loop” description of its AI content strategy with DCN member data on falling traffic, as reported by PPC Land.

What does DCN’s brief say about AI and publisher traffic?

DCN says data from participating members shows a 26% fall in unique visitors between 2023 and 2025. It also reports a median 22% year-over-year drop in referred web traffic over an eight-week window in July 2026, against 10% for the same window in 2025, and a 5% decline in both advertising and total revenue in 2025. PPC Land notes that these survey figures were compiled by DCN itself, and that the judge has not yet decided whether to admit the brief.

The brief’s core argument is that retrieval-augmented generation (RAG) lets AI answers substitute for the articles they draw on, which erodes the revenue that funds the journalism.

Does the data prove AI is causing the decline?

No. This is advocacy from a trade body whose member, the NYT, is a plaintiff. According to PPC Land, the brief does not say how many members responded, which analytics systems were used, or how AI effects were separated from other shifts in search and social distribution. PPC Land also points out that the 2025 figure had been published before, despite the brief’s claim that it had not, and that the brief has inconsistent citations.

The brief also relies on studies of Google, not ChatGPT or Copilot, and argues the defendants’ tools “almost certainly have a similar effect.” That is an inference, not a measurement. Microsoft’s internal Bing Chat click-through comparison, cited from the plaintiffs’ filing, is the strongest number, with reductions of up to 87-93% for NYT sites.

What does the doom loop argument mean for SEO and content teams?

It signals where publisher strategy is heading: not demanding traffic, but demanding payment when content feeds an AI answer. For SEOs, that means licensing and citation, not only rankings, may become part of how content value is measured.

It also shows how weak the click data is. PPC Land notes that Bing’s AI Performance dashboard counts citations in Copilot but reports no click data. If you depend on organic traffic, you are mostly measuring from your own analytics.

One concrete date matters. PPC Land reports that Microsoft said on October 30 that Bing’s NoArchive and NoCache directives will stop acting as grounding controls on March 1, 2027, when separate AI training and AI grounding preferences take over.

How should you respond to AI traffic loss now?

• Segment your own referrals. Compare year-over-year organic and AI-referred sessions for the same period, as DCN did, and note which pages lose most.
• Check your revenue mix. DCN members drew 84% of digital revenue from advertising in 2025; if you are similarly exposed, model a drop in page views.
• Audit Bing directives. If you use NoArchive or NoCache to limit Copilot grounding, plan for the March 1, 2027 change.
• Don’t quote the numbers as industry-wide. Treat them as one group’s survey results when briefing clients.
• Watch the docket. The ruling on the brief and on the fourth fair use factor could shape licensing deals.

Sources

• PPC Land: Publisher group says AI cut visitors 26%, quoting Microsoft’s ‘doom loop’