Britannica and Merriam-Webster Sue Perplexity Over AI Copying

Britannica and Merriam-Webster sue Perplexity

Encyclopaedia Britannica, which owns Merriam‑Webster, filed a federal lawsuit against AI search startup Perplexity on September 10th, alleging copyright and trademark infringement. The complaint points to near‑verbatim reproductions of dictionary entries—most notably Merriam‑Webster’s definition of the word “plagiarize”—and accuses Perplexity of presenting those reproductions without permission or proper attribution.

Beyond copying, Britannica claims trademark harm when Perplexity attaches its and Merriam‑Webster’s names to answers that are incomplete or hallucinated, eroding the trust associated with those brands. The suit also alleges "stealth crawling"—accessing content despite crawler blockers on publisher sites.

Perplexity markets itself as an alternative to Google Search and has faced similar disputes with several media organizations, including News Corp, Forbes, The New York Times and the BBC. The company is backed by notable investors, which raises the stakes for publishers seeking remedies or licensing deals.

• Allegation: scraping protected content and reproducing it in AI answers
• Allegation: trademark misuse when trusted names are shown alongside inaccurate output
• Allegation: bypassing crawler restrictions, sometimes called stealth crawling

This dispute is part of a broader clash between publishers and AI companies over where training and answer content comes from, and who should be paid or credited. Some publishers have struck deals—Time and the Los Angeles Times joined Perplexity’s ad‑revenue sharing, and World History Encyclopedia launched a Perplexity‑powered chatbot—while others have pursued litigation.

Why does it matter? When an AI answer mirrors a copyrighted entry word‑for‑word, the legal, commercial, and trust implications are immediate. Publishers risk losing traffic and licensing revenue. AI platforms risk injunctions, damages, and reputational harm. Consumers lose a dependable signal about what sources are authoritative.

Practical steps for organizations

Publishers, AI builders, and policy teams can take concrete actions now to reduce risk and restore trust. Think of it like maintaining a library: if someone scans pages and hands out photocopies without permission, the librarian needs a record of what was copied and controls on further distribution.

• Audit and provenance: map which external sources train or feed your answers and detect verbatim overlaps
• Clear crawling policies: enforce robots.txt and use crawl detection to stop stealth scraping
• Citation and UX design: surface sources and confidence levels alongside answers to prevent brand‑name misuse
• Commercial paths: negotiate licensing or revenue‑share arrangements where appropriate

For AI firms, designing models and answer engines that default to verifiable citations and that label uncertain content reduces both legal exposure and user harm. For publishers, pairing technical defenses with commercial pathways creates options beyond litigation.

At a systems level, the dispute raises a broader question: who funds reliable knowledge in an AI era? Litigation will shape short‑term behavior, but sustainable outcomes require technical transparency, publisher participation, and business models that recognize content creators’ value.

QuarkyByte’s approach is to blend forensic data analysis with operational policy: trace where answers come from, quantify overlap with protected works, detect stealth crawling, and surface governance steps that reduce legal and reputational risk. Whether you represent a publisher defending original content or an AI team building citation‑first answers, targeted audits and measurable controls make the difference between recurring lawsuits and a stable content ecosystem.

This case will be a bellwether. Expect more litigation and more licensing deals as publishers and AI platforms test where legal lines and commercial incentives meet. The immediate takeaway: transparency and provenance aren't optional—they're the baseline for credible AI answers.