Pre-publish scan for staff reporters, pre-handoff scan for freelancers, and a defensible audit log for editors running a desk policy. Source-quote aware so block quotes from interviews and wire copy do not pollute the score on the reporter prose. Sentence-level highlights with perplexity and burstiness signals so an editor can read the diagnostic, not just the headline number. Free to try. No card.
Staff reporters filing for a desk under a published AI policy, freelance journalists handing copy to editors at multiple outlets, and editorial teams running pre-publish review all share one need: a fast scan that separates source-quote material from reporter prose and points to the specific paragraphs where authorship is in question.
Journalism sits at the intersection of two pressures that make detection harder than in other writing genres. The first is the heavy use of quoted material: a feature with twelve quotes from sources is not the same as a blog post with twelve quotes from research papers. The second is the rhetorical density of opinion and long-form work, which overlaps with the patterns AI models default to. The realistic 2026 workflow scans the right thing in the right context.
Three to ten pieces a week across breaking news, daily news, and weekly features. Pre-publish scanning catches both genuine AI residue from wire-copy rewrites and the false-positive patterns that rhetorical news writing sometimes triggers. Pro at $14.99 a month yearly handles unlimited scans and the 10,000 character cap that covers most news stories in a single paste.
Pitching multiple outlets, often with the same source material reshaped for different editors. Pre-pitch scanning before the pitch goes out, then pre-handoff scanning before the final draft lands in the editor inbox. 90-day scan history on Pro covers the typical pitch-draft-revise cycle for a feature commission.
Five to twenty reporters filing across a desk. Business at $29.99 a month yearly unlocks five seats with shared scan history, REST API access for CMS integration, an audit log that shows which editor scanned which story with timestamps, and white-label PDF exports for archiving alongside the published version. Newsroom standards leads running a minimum score policy land on Business within the first quarter.
Newsroom standards documents at the major English-language outlets have rewritten their guidance on AI use during 2024 and 2025. Pre-publish scanning helps editors verify the disclosure label that goes on a piece, and it gives reporters a defence when a piece labelled fully reported is later challenged.
All six major English-language outlets updated their AI guidance during 2024 and 2025. The common thread is disclosure: AI use in drafting, research, or rewriting should be labelled on the published piece, with the specific tool and the specific use case identified. Pre-publish scanning gives the standards editor a quick way to verify that a piece labelled as fully reported in fact reads as fully reported, and to flag pieces where the disclosure may need to be expanded.
The audit log on Business is the piece that makes scanning fit a newsroom policy rather than sitting as an individual tool. Reporters file, editors scan, and the log records who scanned what and when. A standards review later in the year can pull the log to demonstrate consistent quality control across the desk. A challenge from outside the newsroom on a specific piece can be answered with the contemporaneous scan record.
The two are different jobs. Disclosure is editorial policy: did the reporter use an AI tool in the drafting workflow, and is that labelled on the piece. Detection is technical: does the published prose read AI-shaped to a classifier. A piece can be fully reported and still score in the templated band because rhetorical news writing overlaps with AI patterns. The right read is to use the scan as one signal among the editorial judgements an editor already makes, not as the final word on authorship.
Pro at $19.99 a month standard, $14.99 a month on yearly, is the right fit for freelance journalists filing across multiple outlets. Business at $39.99 a month standard, $29.99 a month on yearly, fits newsrooms and editorial teams running pre-publish scans across a desk. Full details on the pricing page.
Billed $89.88/year — Save $30
Billed $179.88/year — Save $60
Billed $359.88/year — Save $120
Yearly billing saves 25%. View full pricing →
The scan slots into the editorial cycle in two specific places: pre-publish for staff reporters once the desk editor approves the draft, and pre-handoff for freelancers before the final draft moves from the reporter to the commissioning editor. Same tool, two different gates.
After the desk editor approves the draft and before the piece moves to copy edit. The point is to surface any AI-shaped prose that crept in during the rewriting loop, and to give the standards editor a contemporaneous record of the scan score and the timestamp. On Business the audit log captures all of this automatically. On Pro the PDF export does the same job for a freelance editor working without team infrastructure.
Before the final draft leaves the reporter and lands in the editor inbox. The freelancer scans, reviews the sentence highlights, and rewrites any paragraphs running below the desk floor. The point is to avoid the back-and-forth where an editor runs a third-party detector, gets a flagged result, and the freelancer has to defend prose they wrote themselves. A clean pre-handoff scan attached to the delivery email pre-empts that conversation.
Desk editors running a minimum score policy across a team need to see the pattern, not just individual scans. On Business the audit log shows scan history across all five seats with reporter name, story slug, score, and timestamp. Useful for spotting the reporter whose pieces consistently sit at the floor, the desk where the rhetorical opinion work flags more than the news copy, or the assignment cycles where revision is dropping the score rather than lifting it.
If an editor requests substantive edits, rescan the revised draft before it goes back through. This catches the case where a revision accidentally flattens a paragraph while fixing a different issue. Keeping a contemporaneous record across the revision cycle prevents a late-stage challenge on a piece that scanned clean originally.
A feature with twelve quotes from sources is structurally different from a blog post with twelve quotes from research papers. Quoted material from interviews is not authored prose in the same sense, and the scan should be read in that context. The result view makes this explicit at the sentence level.
A direct quote pulled verbatim from a recorded interview is not reporter prose and should not weigh into the authorship question. The TextSight result view colour-codes every sentence, and quoted material reads with its own pattern: short, conversational, often grammatically rough in a way reporter prose rarely is. An editor reviewing the scan can ignore those rows and focus on the narrative paragraphs where the authorship question actually lives.
For long-form features with substantial block-quoted material, exclude the quote blocks from the calibration paste. Scan the reporter prose alone first, then scan the full piece with quotes included for the headline number that goes on the desk record. The two-pass approach separates the diagnostic (did I write this human prose) from the disclosure (what does the published piece score in full).
Wire-service material rewritten by a reporter is the reverse case: it is published as reporter prose but the underlying language comes from AP or Reuters. Rewrites that strip out the wire-service voice and add reporter context tend to score normally. Rewrites that leave the wire voice intact and only change a few words tend to flag. Pre-publish scanning catches the second pattern before it ships.
A reporter using an AI tool to summarise interview transcripts is using a research aid, not generating publishable prose. The notes are not the deliverable; the article is. Scan the article, not the notes. The disclosure question lives at the editorial policy layer, not the detection layer, and a reporter using AI for transcription is making a different editorial choice from a reporter using AI for drafting.
Genre matters more in journalism than in most writing markets. Straight news copy scores high on average because the prose is concrete and reportorial. Opinion columns, long-form features, and investigative pieces score lower because the prose is denser and more rhetorical, and rhetorical patterns overlap with AI training data.
Eight-hundred to fifteen-hundred word news stories with sourced quotes, datelines, and concrete reporting detail. Healthy scores run 80 to 95 on prose that comes from real reporting. The intro paragraph is the highest-risk part of a news story because news intros default to a templated lede structure. Read the sentence highlights and rewrite the intro if it sits below the desk floor.
Eight-hundred to twelve-hundred words of rhetorical argument. Healthy scores run 65 to 80. Opinion writing flags more often because its rhetorical patterns overlap with AI training data, not because the writer used AI. Read the sentence highlights rather than the headline number. Target the templated transitions and the editorial summaries; defend the columnist voice on its merits.
Three-thousand to ten-thousand words with multi-source reporting, narrative arcs, and substantial block-quoted material. Healthy scores run 70 to 85 on the reporter prose component. Scan in two passes for long features: the prose alone for the diagnostic, and the full piece for the disclosure record. The intro section and the closer are the highest-risk segments. Read the paragraph cards on Pro to find the lowest-scoring section first.
Multi-month investigations with documentary evidence, recorded interviews, and dense source attribution. Tend to score in a wide band because the prose ranges from rhetorical (the framing sections) to concrete (the documentary sections). Read the paragraph cards rather than the headline. The audit log on Business is particularly useful for investigations because the standards review later in the year usually wants the contemporaneous scan record across every draft.
A single percentage is not a fix path for a desk editor. The TextSight result panel shows which sentences reacted and why, with paragraph-level rollups for longer pieces, so a reporter or editor can target specific lines instead of rewriting the whole story.
Every sentence is colour-coded by its own AI-likeness score. Red sentences clustered in one paragraph are a stronger signal than scattered yellows. Scattered yellows in otherwise structured prose often just mean a stock transitional phrase. An editor reads the pattern across the piece, not the headline number on a sticker.
Longer pieces get paragraph-level rollups so a reporter can see which section is dragging the headline score. The intro and the closer are the usual suspects on news copy, while the rhetorical transitions are the usual suspects on opinion work. Targeting the lowest paragraph first is the fastest way to lift the story.
Perplexity is how predictable your word choices are to a language model. Low perplexity reads AI-like. The score is shown per-sentence on Pro, which gives a reporter the diagnostic context to decide whether a flag is real AI residue or a particularly well-rehearsed news-style sentence.
Burstiness is how much sentence length and structure vary across the piece. ChatGPT defaults to uniform medium-length sentences. Real reporting has bursty rhythm: one short sentence, one long, one fragment, a quote pull. Low burstiness across an entire story is the classic AI fingerprint and the one desk editors learn to spot first.
More for journalists.
The pre-deliver scan workflow for blog posts, landing pages, and brand content.
For writers →Light, Balanced, and Maximum modes for fixing flagged passages without losing voice.
Read the guide →What the headline number means and how to read paragraph cards on Pro.
See the score →Free, Starter, Pro, Business. Yearly billing saves 25%. Freelance to newsroom tiers.
See pricing →Free to try. No card. Pro at $14.99 a month on yearly for freelance reporters; Business at $29.99 a month on yearly for newsrooms and editorial teams.