{ "source": "Cooked Index — occupational AI risk register", "page": "https://cookedindex.com/jobs/news-analysts-reporters-and-journalists/", "methodology": "https://cookedindex.com/methodology", "notice": "Verdicts are re-examined as evidence accumulates. Re-fetch before relying on this; the page above always carries the current score.", "scored_at": "2026-08-11", "model": "claude-opus-5", "occupation": { "title": "News Analysts, Reporters, and Journalists", "soc_code": "27-3023", "category": "Media", "us_employment": 39250, "median_annual_wage": 62200 }, "verdict": "EXPOSED", "risk_resistance": 37, "contested": false, "near_boundary": false, "dimensions": { "task_resistance": 7, "embodiment": 8, "liability_shield": 1, "trust_premium": 10, "judgment_accountability": 11 }, "reasoning": { "task_resistance": "A 7 rather than a 12 reflects how much of the week is text-transformation the model already handles — turning a police blotter, an agenda packet, a quarterly filing or an AP feed into 400 publishable words — with only the doorknocking, the off-record coffee, and the FOIA-follow-up sitting above that line.", "embodiment": "An 8 covers the real but intermittent physical duty: standing at the school board meeting, driving to the flood, holding a recorder at a scrum, shooting your own phone video — none of it in the uncontrolled or hazardous conditions a war correspondent or storm-chaser faces, and much of the day still filed from a laptop.", "liability_shield": "A 1, not a 0, because journalism is deliberately unlicensed in the US — the First Amendment forecloses credentialing — so nothing but a press pass anyone can print stands between the job and a competitor, human or otherwise; defamation exposure lands on the publisher, not on you personally.", "trust_premium": "A 10 splits the beat in two: the sources who leak to you specifically because they know you burned nobody before are irreplaceable, but the reader who arrives from a search result or an aggregator has no idea who wrote the story and would not change habits if the byline vanished.", "judgment_accountability": "An 11 is right for calls like whether two sources are enough to publish, whether to name a juvenile, what to withhold in an active manhunt, and where the line sits between allegation and fact — genuine discretion, but exercised inside a style guide and cleared by an editor who takes the final publish decision." }, "rationale": "The modal worker here is a local or trade reporter whose day mixes original reporting with a large volume of commodity text: rewriting press releases, recapping earnings and game results, summarizing public meetings, aggregating wire copy, and writing SEO headlines — all of which generative models already do at publishable-enough quality for ad-supported outlets. What resists is the part that requires being in the room: cultivating sources who only talk to a person they trust, showing up at the courthouse or the fire, verifying a claim against a document nobody has posted, and personally owning a story that could get the outlet sued. There is no license protecting this work, and the economics of the industry are shrinking headcount independent of AI, which compounds the exposure.", "outlook": "By the mid-2030s commodity news writing is largely machine-generated and staff counts keep falling, while a smaller tier of source-driven, on-scene, and investigative reporters with their own audiences becomes more valuable per person.", "what_would_raise_it": { "levers": [ { "dimension": "task_resistance", "change": "Commodity tier (press-release rewrites, earnings/game recaps, meeting summaries, SEO heds) fully absorbed by newsroom AI stacks, leaving surviving headcount concentrated on source-based original reporting and document verification — a genuine two-tier occupation where the residual job is the judgment tier. Watch AP/Gannett/Patch automation rollouts and whether remaining job postings specify 'investigative' or 'accountability' beats.", "plausibility": "already happening", "would_add": 4 }, { "dimension": "liability_shield", "change": "A named-human accountability requirement rather than a license: EU AI Act transparency obligations plus platform/defamation exposure pushing outlets to require a credentialed human byline who personally attests to verification before publication. Also watch insurers — media liability carriers (e.g. Axis, Chubb media E&O) adding policy conditions that AI-generated copy must be signed off by an identified editorial employee, and NYT v. OpenAI-style litigation making unattested synthetic copy uninsurable.", "plausibility": "plausible", "would_add": 4 }, { "dimension": "trust_premium", "change": "Provenance labeling becoming a purchasable signal: C2PA/Content Credentials adoption plus 'human-reported' disclosure norms (already in AP and Reuters standards, and in some Writers Guild/NewsGuild contract language) so subscribers and syndication buyers pay specifically for verified-human bylines. Watch whether Google/Apple News surfaces or licensing deals price human-reported content differently, and whether NewsGuild contracts win AI-disclosure and byline-consent clauses.", "plausibility": "plausible", "would_add": 4 }, { "dimension": "judgment_accountability", "change": "If courts continue to treat the named reporter/editor as the actual-malice locus, and if shield-law protections (federal PRESS Act as introduced) attach only to identifiable human journalists who can be subpoenaed and can refuse, the consequential call — publish or hold, name or protect a source — stays personally owned and cannot be delegated to a model.", "plausibility": "plausible", "would_add": 3 } ], "ceiling_note": "Even with every lever, headcount economics dominate: the score describes per-role resistance, not the number of roles, and local ad-supported outlets are shedding jobs regardless. No licensure route exists for journalism in the US — First Amendment doctrine effectively forbids it — so liability_shield cannot reach the levels seen in law or medicine." }, "adjudication": null, "employment_history": { "points": [ { "y": 2019, "emp": 44100, "wage": 46270 }, { "y": 2020, "emp": 41580, "wage": 49300 }, { "y": 2021, "emp": 39080, "wage": 48370 }, { "y": 2022, "emp": 44530, "wage": 55960 }, { "y": 2023, "emp": 45020, "wage": 57500 }, { "y": 2024, "emp": 41550, "wage": 60280 }, { "y": 2025, "emp": 39250, "wage": 62200 } ], "from": 2019, "to": 2025, "change_pct": -11, "comparable_from": 2019, "spans_soc_revision": false }, "pivots": [ { "slug": "detectives-and-criminal-investigators", "title": "Detectives and Criminal Investigators", "verdict": "SAFE", "risk_resistance": 69, "median_wage": 93790, "overlap": 73, "skills_to_close": [ "Science", "Critical Thinking", "Troubleshooting", "Operations Monitoring" ] }, { "slug": "police-and-sheriff-s-patrol-officers", "title": "Police and Sheriff's Patrol Officers", "verdict": "SAFE", "risk_resistance": 82, "median_wage": 76210, "overlap": 65, "skills_to_close": [ "Operations Monitoring", "Troubleshooting", "Science", "Operation and Control" ] }, { "slug": "history-teachers-postsecondary", "title": "History Teachers, Postsecondary", "verdict": "EXPOSED", "risk_resistance": 47, "median_wage": 83820, "overlap": 65, "skills_to_close": [ "Learning Strategies", "Instructing", "Operations Analysis", "Science" ] } ], "license": "https://cookedindex.com/terms" }