New York Times Sues Perplexity AI Over Copyright Infringement
The New York Times escalates its legal campaign against artificial intelligence firms by targeting Perplexity AI, alleging the startup systematically copies and redistributes millions of its articles without authorization. This lawsuit, filed in the U.S. District Court for the Southern District of New York, accuses Perplexity of violating copyrights through its generative search engine, which scrapes paywalled content to produce near-verbatim outputs. The action arrives as Perplexity faces parallel suits from publishers including the Chicago Tribune and Dow Jones, underscoring fractures in the AI ecosystem where content fuels model training but erodes creator revenues.
Perplexity, a San Francisco-based company founded in 2022 by Aravind Srinivas and three former OpenAI and Google engineers, operates a conversational search platform valued at $20 billion following a $200 million funding round in September. The startup’s core technology relies on retrieval-augmented generation, where large language models query indexed web data—including Times articles, videos, and podcasts—to formulate responses. According to the complaint, these outputs substitute for direct visits to the Times website, diverting subscription fees, advertising impressions, and licensing opportunities estimated in the tens of millions annually.
The Times first issued cease-and-desist notices to Perplexity in July 2024 and November 2024, demanding negotiations for licensed access, but the startup continued its practices. The suit details instances where Perplexity reproduced full articles verbatim or generated summaries with fabricated attributions, such as falsely crediting the Times for non-existent stories. It seeks unspecified damages, an injunction to halt further use, and destruction of infringing data caches, joining over 40 similar cases against AI developers.
Perplexity’s defenders argue its methods constitute fair use under U.S. copyright law, transforming scraped content into novel analyses rather than direct copies. The company, which raised $1 billion total from investors including Jeff Bezos and Nvidia, emphasizes ethical sourcing and has inked deals with outlets like Time for voluntary partnerships. Yet the Times counters that Perplexity’s commercial model—generating $100 million in annual revenue—directly competes with journalism by offering free alternatives that undermine paywalls.
Graham James, a Times spokesperson, stated in the filing announcement: “While we believe in the ethical and responsible use and development of AI, we firmly object to Perplexity’s unlicensed use of our content to develop and promote their products.” Perplexity did not immediately respond to the suit but has previously dismissed similar claims as misguided attempts to stifle innovation. The case builds on the Times’ 2023 lawsuit against OpenAI and Microsoft, where it alleged training on 4 million articles without permission, a dispute now in discovery.
This litigation highlights a pivotal tension in AI’s growth trajectory: publishers demand compensation for data that powers models processing 100 billion queries yearly, while startups prioritize open-web access to compete with Google Search. Perplexity’s tools, including the Comet browser extension and enterprise APIs, integrate with 10 million monthly users, amplifying the stakes. Legal experts predict settlements could standardize licensing fees at 0.5% to 2% of AI firm revenues, potentially reshaping content marketplaces.
For U.S. media firms, already grappling with 20% audience declines since 2020, these suits represent a bid to reclaim value from AI intermediaries. The Times reported $2.4 billion in 2024 revenue, with digital subscriptions comprising 60%, but warns unchecked scraping could halve that stream. As federal courts consolidate related actions, outcomes may influence broader regulations, including the proposed AI Content Transparency Act mandating disclosure of training datasets.
Perplexity’s rapid ascent—from seed stage to unicorn in two years—relies on aggressive data ingestion, but mounting suits risk operational disruptions. The startup recently expanded to Europe with GDPR-compliant features, yet U.S. rulings could impose retroactive audits. This clash tests whether AI’s efficiency gains justify collateral damage to information ecosystems, with implications for 5,000 U.S. newsrooms facing existential threats from automated summarization.
