Use-case guide

AI in Journalism & News Media

News organisations face AI-driven exposure on three fronts at once: as content producers (AI-drafted articles, AI-generated images), as platforms (recommendation algorithms, comment moderation), and as litigation defendants (the NYT/OpenAI case is the most visible, but others are pending). The regulatory load is lighter than in heavily-regulated sectors but the reputational and defamation exposure is unusually high — a hallucinated fact in a CNN/Reuters/AP byline is a Section 230 / European Convention Article 10 catastrophe in a way it isn't on a personal blog.

For: Editors-in-chief, news platform tech leads, journalism standards officers, syndicators, fact-check organisations, media defense counsel

What's at stake

Hallucinated facts create defamation exposure

Walters v. OpenAI (Georgia) and parallel cases are testing whether ChatGPT-style hallucination of false facts about a real person establishes defamation. Publishing organisations face higher actual-malice-equivalent scrutiny.

EU AI Act + DSA transparency obligations stack

AI-generated news content must be labelled under EU AI Act Article 50. DSA Article 39 requires VLOPs (and most major news platforms qualify) to disclose ad-targeting and recommendation logic.

Source-verification duties of care under journalistic standards

IFCN code, Reuters Trust Principles, AP Standards, BBC Editorial Guidelines all now incorporate AI-tool standards. Failure to apply standard verification to AI-sourced content is treated as standards violation, not just AI-policy violation.

Training-data copyright cases reshape licensing

NYT v. OpenAI and parallel cases are catalysing direct AI-content-licensing deals (NewsCorp/OpenAI, AP/OpenAI, FT/OpenAI). Publishers without a licensing posture face revenue erosion + IP exposure.

Regulations that apply

EU AI Act

Law
→

Article 50 labelling for AI-generated content (including news); Article 52 deepfake disclosure; Article 53 GPAI duties for foundation models used by news orgs.

Where in the text: Articles 50, 52, 53.

EU Digital Services Act

Law
→

News platforms that qualify as VLOPs/VLOSEs have systemic-risk assessment + audit obligations covering AI-driven recommendation + content-moderation. Non-VLOP news orgs still have transparency duties.

Where in the text: DSA Articles 26, 34, 35, 37, 39.

First Amendment + actual-malice defamation

Law
→

US press faces NYT v. Sullivan actual-malice standard for public figures. AI-hallucinated facts about public figures test whether reckless-disregard threshold is met.

Where in the text: NYT v. Sullivan 376 U.S. 254 (1964); Gertz v. Welch 418 U.S. 323 (1974).

ECHR Article 10 + national defamation law (EU)

Law
→

European Convention Article 10 free-expression interacts with national defamation regimes. UK Defamation Act 2013 + EU Member State laws apply to AI-generated false statements.

Where in the text: ECHR Article 10; UK Defamation Act 2013; comparable national codes.

Do

  • ✓Treat any AI-generated content destined for publication with the same verification standards as a stringer's filing — name + date + quote validation; primary-source review.
  • ✓Label AI use in bylines + article metadata. The Reuters / AP / Guardian model (clearly disclosed) is becoming the industry floor.
  • ✓Document your fact-check methodology for AI-drafted content. Defamation defense leans heavily on documented verification process.
  • ✓Get a corrections-policy AI-extension drafted — most existing corrections policies pre-date AI hallucination risk and don't fit the failure mode.
  • ✓License training-data carefully when contracting AI assistance vendors. The 'all my archive for $X' model trades a fixed payment for unknown downstream exposure.

Don't

  • ✗Don't publish AI-drafted quotes attributed to real people without independent verification — this is the Walters v. OpenAI pattern but with publisher amplification.
  • ✗Don't deploy synthetic 'AI anchors' or AI-generated reporter avatars without Article 50 disclosure in EU markets and clear visual cues elsewhere.
  • ✗Don't use AI to generate user-comment moderation decisions without an appeal route — DSA Article 17 requires statement-of-reasons for moderation actions.
  • ✗Don't omit the AI-tool in your sources list when an AI tool sourced a fact. The NYT and FT standards now treat the AI tool as a named source.
  • ✗Don't reuse AI-generated images without checking provenance (C2PA, SynthID). Photo desks face increasing scrutiny over AI-generated visual content.

Also worth knowing

For news platforms with comment sections: AI-driven moderation falls under DSA + Section 230 differently across jurisdictions; document the human-in-the-loop. For data-driven journalism: AI used in investigative work has unique source-protection considerations — particularly around model prompts that include source-quoted material. For wire services (AP, Reuters, AFP): contractual flow-down of AI use disclosure is now standard in syndication contracts.

Want a tailored answer?

The wizard takes your jurisdiction, AI use case, and data types and gives you the top-3 regulations to focus on — in 60 seconds.

Start the wizard →

Educational guide. Not legal advice. For specific compliance decisions, consult qualified counsel in the relevant jurisdiction.

Note: this guide was drafted with AI assistance — Anthropic Claude.