82% of journalists now use some form of AI in their work, according to Muck Rack's 2026 State of Journalism report. Meanwhile, Google search traffic to publishers has declined by a third in the past year, driven by AI Overviews replacing click-throughs.
The Big Picture
Reuters Institute Predictions 2026: The Scorecard
The Reuters Institute surveyed 280 news executives across 51 countries. Their verdict: AI is shifting from experiment to core infrastructure in newsrooms. Back-end automation was rated important by 97% of respondents. Search referral traffic to publishers fell a third in 2025, and publishers forecast another 43% decline within three years. Breaking news, counterintuitively, is up 103% across Google surfaces — the one format AI summaries haven't eaten yet. The report flags agentic AI (multi-step autonomous workflows) as the next frontier, though most newsrooms aren't there yet.
Industry Trends
AI and the Future of News 2026: What We Learnt
Recap of the Reuters Institute's annual conference. Key themes: AI coverage itself often lacks expert sources and defaults to tech company narratives. Climate journalist Akshat Rathi draws parallels — the same companies that abandoned climate commitments have pivoted to AI hype. Panelists urge journalists to demystify AI by collaborating with academic researchers and being precise about which "AI" they mean. The Guardian shared its approach to integrating AI while maintaining editorial standards.
Coverage & Ethics
Artificial Intelligence in Journalism — Issue Primer
A comprehensive policy primer. Generative AI offers newsrooms productivity gains in summaries, local event coverage, and comment moderation, but comes with risks: inaccuracies, copyright conflicts, erosion of public trust. The report argues transparency alone isn't enough — the public lacks context about journalistic practices to interpret AI disclosures meaningfully. Calls for forward-thinking collaboration among policymakers, publishers, and tech developers.
Policy
In the Newsroom
AI Advice from Journalists Who Stopped Talking and Started Building
Five lessons from SXSW's AI x Journalism Day. Start with a specific pain point, not a mandate — Pew Research built a WordPress plugin to auto-draft formulaic social posts, freeing their team to actually engage audiences. Nebraska Public Media stopped using AI for writing entirely but calls it a "good curiosity partner" for research. Texas Tribune's AI chatbot, trained on school voucher coverage, surfaced reader questions that led to new stories. The consensus: draw a clear line between AI for thinking and AI for writing.
Workflow
Meet the Tech Reporters Using AI to Write and Edit Their Stories
Independent journalist Alex Heath has integrated Claude into his entire workflow — connected to Gmail, calendar, transcription, and notes. He speaks his ideas into a mic, lets the AI draft, then iterates for 30 minutes. Says it cuts his writing time by 30-40%. The trend is especially strong among solo reporters who've lost access to traditional newsroom resources like editors and fact-checkers. They're using AI to recreate those roles rather than replace the journalism itself.
Tools
How AI Agents Are Changing Journalism
The rise of AI agents could automate much of newsroom operations — production, distribution, even some editorial triage — leaving human journalists to focus on taste, judgment, and trust. The piece argues the value of a journalist is increasingly about what machines can't do: building source relationships, making ethical calls, and earning reader trust.
AI Agents
Risks & Challenges
The First AI Election and the Newsroom Infrastructure Gap
If Iran was the first AI war, November 2026 is the first AI election. AI-cloned candidate voices and synthetic campaign ads are already running without meaningful disclosure. Local newsrooms — the frontline for midterm coverage — lack the verification infrastructure to handle the volume of AI-generated content. The author warns against two bad instincts: abstention (stepping back from contested races) and over-centralization (letting AP handle everything). Both leave dangerous gaps.
Misinformation
How People Feel About AI in Journalism
A seven-country study on public attitudes toward AI in newsrooms. The takeaway: AI might have a place in journalism, but everyone should proceed with caution. Audiences are more accepting of AI for behind-the-scenes tasks (research, transcription) than for content creation. Trust declines sharply when readers learn a story was AI-generated, even if the quality is comparable.
Public Trust
IP & Copyright Wars
Music Publishers Make Their Case Against Anthropic's Fair Use Defense
Universal, Concord, and ABKCO are suing Anthropic for $3.1 billion, alleging Claude was trained on torrented pirated music. The publishers asked the court to reject Anthropic's fair use defense before trial. Meanwhile, a judge ordered OpenAI to produce all 20 million output logs in the NYT case. The September 2026 trial dates are approaching — these rulings will set precedent for whether AI training on copyrighted work qualifies as fair use.
Copyright
Courts Begin to Draw Lines Around AI Training and Market Harm
2025 saw courts start to establish boundaries. Key question emerging: are model weights "copies" of training data? Defendants argue no; plaintiffs say the economic harm is the same regardless. Ziff Davis (IGN, Mashable, CNET, PCMag) is suing OpenAI for scraping 45+ digital publications. The cases are converging on a core tension: does the output market substitute for the original, or is it transformative enough?
Copyright
Regulation & Free Speech
Court Cases Bypass Section 230 as AI Generates Its Own Content
A wave of lawsuits is testing whether Section 230 protects platforms when AI generates content rather than users. The legal theory: when Google's AI Mode or Meta's AI features create original responses, the platform is no longer a passive host — it's an author. Courts haven't fully explored this yet, and the outcomes could fundamentally reshape platform liability. The timing collides with Senator Blackburn's TRUMP AMERICA AI Act, which proposes repealing Section 230 entirely.
Section 230
EFF: Generative AI and Section 230's Grey Area
The EFF maps the spectrum: user-generated content hosted by a platform (clearly protected) to content the platform "materially contributed to" in harmful ways (clearly not). Generative AI sits in the middle. The policy implications cut both ways — too much liability could kill AI innovation; too little could remove any accountability for AI-generated harms. For newsrooms, this matters because AI-assisted articles may shift legal exposure from individual reporters to the publication itself.
Policy
White House Unveils National AI Framework
The March 20 framework proposes federal preemption of state AI laws and explicitly states AI should not be used by government to suppress lawful expression. No new federal rulemaking body — the administration prefers existing regulators and industry standards. The libertarian read: a relatively light touch, but the preemption angle could prevent states from experimenting with stronger protections. The free speech language is notable but untested.
Free Speech
Deepfakes & the 2026 Midterms
AI Deepfakes Blur Reality in 2026 Midterm Campaigns
The NRSC released an AI-generated video of a Texas candidate reciting his own old social media posts — with "AI generated" in barely visible fine print. Campaigns on both sides are already deploying synthetic ads. Deepfake attempts rose 303% around recent U.S. elections (Sumsub data). The gap between what campaigns can produce and what newsrooms can verify is widening fast, and the midterms are seven months out.
Deepfakes
How Cognitive Manipulation and AI Will Shape Disinformation in 2026
2026 is packed with elections across continents. The WEF report warns that the speed and scale of synthetic media is a "compounding risk" — each cycle, the tools get better and cheaper while detection lags behind. The most vulnerable targets are local newsrooms covering state and congressional races, which lack the verification infrastructure of national outlets.
Elections
AI & Documentary Filmmaking
"The AI Doc" Gets Chief Executives on the Record
Daniel Roher (Oscar winner for "Navalny") and Charlie Tyrell's new documentary features interviews with Sam Altman, Dario and Daniela Amodei, and Emily Bender. Currently 8.2 on IMDb. The film takes the approach of a wondering inquisitor rather than a polemic — asking what's happening rather than declaring what should. For documentary journalists, it raises the question: how do you tell the AI story when the subject is moving faster than any production timeline?
Documentary
An AI Company Set Out to Fix News Deserts. Instead, It Copied Local Journalists' Work.
Nota News promised to fill local news gaps with AI-generated coverage. Instead, it scraped and rephrased work from the same struggling local outlets it claimed to supplement. The story captures the fundamental tension: AI can theoretically expand news coverage into underserved areas, but the business incentives point toward parasitism rather than creation. The economics of local news are already broken — AI is accelerating the damage faster than it's producing solutions.
Local News
Key Themes to Watch
What's emerging
- Search traffic collapse is forcing newsrooms to invest in direct audience relationships — newsletters, apps, community
- AI licensing deals between publishers and AI companies are a potential new revenue stream, but the economics remain uncertain
- Agentic AI (autonomous multi-step workflows) is the next wave — most newsrooms are building the infrastructure now
- The 2026 midterms will be the first real stress test of AI-driven misinformation at electoral scale
- Independent journalists are adopting AI tools faster than institutional newsrooms, using them to replace editors, fact-checkers, and research assistants
- Public trust is the limiting factor — audiences accept AI for back-office work but resist AI-written content
- AI copyright trials (NYT v. OpenAI, music publishers v. Anthropic) hitting courtrooms in late 2026 — will define fair use for a generation
- Section 230 is being tested from both sides: courts questioning AI-generated content protections, Congress proposing full repeal
- Deepfake political ads are already running in midterm campaigns with minimal disclosure — verification infrastructure doesn't exist at local level
- White House AI framework favors federal preemption over state regulation — lighter touch, but preemption limits experimentation