NYT doubles down: sulzberger slams AI 'theft' as news industry fights back with bans, lawsuits, and coalitions

The New York Times' strict enforcement against AI-generated content reflects deep ethical challenges in journalism, particularly regarding the accuracy and fabrication risks posed by AI tools. Steven Rosenbaum’s experience, where his book contained multiple fabricated or misattributed quotes generated by ChatGPT and Claude, underscores how AI hallucinations can slip through even rigorous fact-checking processes, as he admitted, 'We look at it, it looks right. We double-check it, and then we made a mistake.' This highlights the difficulty news organizations face in fully policing AI inaccuracies while striving to maintain journalistic integrity.

Despite the pitfalls, many journalists continue to embrace AI as a 'delightful writing companion,' as Rosenbaum puts it, even while acknowledging its tendency to 'betray you in ways that are just really quite horrible.' This ongoing reliance amplifies ethical dilemmas around transparency and disclosure, with experts emphasizing that AI-assisted journalism requires strict practices to trace and verify chatbot-generated content before inclusion. Failure to do so risks professional embarrassment and breaches of trust, as careless AI use has led to numerous avoidable errors likened to 'Wile E. Coyote-style' self-inflicted blunders.

The rise of AI-generated news outlets featuring fabricated bylines and no real reporters, such as the South Florida example uncovered by WLRN and WUWF, further complicates ethical standards in journalism. These developments not only undermine traditional notions of authorship and accountability but also exacerbate challenges in transparency and accuracy, as AI-written articles with invented bylines blur the line between legitimate reporting and synthetic content. This trend threatens to erode public trust and calls for more stringent disclosure and verification protocols.

A.G. Sulzberger, publisher of The New York Times, has emerged as a vocal and uncompromising critic of AI companies, accusing them of committing an "original sin" by engaging in what he calls "brazen theft" of intellectual property. He highlights how these tech giants—worth an estimated $11 trillion collectively—have built their AI empires by appropriating journalism, books, music, and film without permission or fair compensation, likening their actions to construction workers building towers from stolen materials. Sulzberger underscores the economic imbalance where AI firms lavishly invest in talent and infrastructure but deliberately withhold fair payment to the creators whose content fuels their systems, with licensing deals reportedly amounting to less than half a percent of the value generated.

Sulzberger sharply criticizes the media industry's fragmented and passive response to AI's exploitation, warning that polite resignation will lead to extinction. At the World Media Congress in Marseille, he urged publishers from over 60 countries to resist Big Tech's content pillaging "by all means—technical, legal, political, and legislative"—and cautioned against negotiating from a position of weakness. He also exposed AI companies' divide-and-conquer tactics, such as OpenAI's preference for selective licensing deals with a few large groups rather than collective bargaining, which weakens the press's negotiating power and risks the media becoming subservient to tech giants as AI increasingly dominates news distribution.

While Sulzberger does not reject negotiations outright, he insists that any licensing agreements must be part of a broader, united front to end predatory practices and secure fair compensation for media creators. He stresses that wanting AI to accurately represent journalism and wanting to be paid for it are aligned goals, yet current deals often pay publishers less than 30% of their content's value, with no transparent market to assess fair pricing. The New York Times itself has invested over $20 million and two and a half years in litigation against AI firms like OpenAI, Microsoft, and Perplexity, positioning itself as the last line of defense against legal precedents that could grant Big Tech unfettered rights to use protected content.

Challenging common justifications AI companies offer for their data appropriation—such as innovation needs or fair use—Sulzberger argues these rationales collapse under scrutiny and warns that dismantling intellectual property protections ultimately harms the creative economy. He invokes Stewart Brand’s lesser-known adage that "Information wants to be expensive," emphasizing the intrinsic value and cost of producing quality content. By framing AI’s foundational act as theft, Sulzberger calls on the media industry worldwide to recognize the economic and ethical stakes and to mobilize decisively against Big Tech’s exploitation.

By early 2026, the AI adoption landscape is marked by sharply rising costs that have prompted major tech firms like Uber, Meta, Microsoft, Salesforce, and DoorDash to implement cost-cutting measures, signaling the definitive end of the 'free-money' era for AI usage. This financial pressure has driven media publishers to seek sustainable revenue streams through six-figure licensing deals facilitated by platforms such as Snowflake’s Cortex, where seventeen publishers—including The Washington Post, Associated Press, and USA Today Network—have legally monetized their proprietary content, circumventing contentious open-web scraping disputes. However, these licensing markets are increasingly dominated by Big Tech companies that once suppressed referral traffic to news sites but now act as gatekeepers, extracting significant fees and setting deal structures that may entrench unfavorable terms for publishers, as highlighted in the Open Markets Institute’s report 'Same Gatekeepers, New Tollbooths.'

Legal confrontations have intensified as news organizations like CNN have initiated landmark copyright infringement lawsuits against AI companies such as Perplexity for unauthorized use of journalistic content, marking a new battleground for fair compensation. The New York Times exemplifies a dual strategy by simultaneously pursuing prolonged litigation against OpenAI and Microsoft while negotiating licensing agreements with Amazon and others, underscoring the high stakes and costly nature of enforcing intellectual property rights in the AI era—cases that can span years and drain resources beyond the reach of smaller publishers. This dynamic reveals a stark economic imbalance where the six leading AI firms, collectively valued at $11 trillion with $350 billion in U.S. private investment in 2025, pay publishers less than half a percent of the value generated from their content.

Publishers are leveraging emerging regulatory and coalition efforts to reclaim bargaining power in AI content licensing. The UK’s Competition and Markets Authority has mandated that Google, which controls approximately 90% of the search market, must allow publishers to opt out of AI-generated content summaries, a groundbreaking shift that disrupts Google's previous 'take-it-or-leave-it' dominance and grants publishers crucial negotiation leverage. Complementing this, the SPUR Coalition unites publishers to establish fair AI licensing standards, with industry leaders like The Guardian’s Doug McCabe projecting that coordinated licensing terms could yield three to four times more revenue for UK journalism over five years compared to fragmented approaches. Yet, strategic differences persist, as exemplified by Le Monde’s CEO Louis Dreyfus, who favors individual licensing deals over collective action, reflecting ongoing debates about the optimal path forward.

The economic stakes of AI integration extend beyond licensing into how AI platforms monetize journalistic content. Google’s Q1 2026 financials, with a 19% increase in Search and 'Other' revenue reaching $89.6 billion, demonstrate that AI-powered products like Conversational Discovery ads and Business Agents for Leads embed advertising directly within AI-generated answers, transforming the value proposition from driving traffic clicks to monetizing the information experience itself. This evolution intensifies the imperative for publishers to secure fair compensation, as their high-quality journalism underpins AI models—The New York Times alone being the largest single proprietary data source among the top ten used in major large language models—yet they currently receive only a fraction of the economic benefits generated.