TheChecker.AI Blog
How AI detection works, what signals to trust, and practical guidance for educators and teams who need content integrity.
Why AI Detectors Struggle Outside English (And What That Means for Your Score)
A detector trained mostly on English can lose 25.7% of its accuracy in other languages. Here's the real research behind that gap.
What Congress's Floor Speeches Reveal About How AI Detectors Spot Machine Writing
A 135-million-word study of Congress shows which vocabulary tells give away AI-written speeches, and why some lawmakers use far more of them.
Should a Single AI-Detector Score Ever Decide a Case?
A fresh CACM feature on a College of Charleston lawsuit shows what actually loses in court: not a wrong score, a missing conversation.
Where AI Writing Actually Concentrates (It's Backwards From What People Guess)
A Rutgers researcher tested 100,000+ documents. People guess AI writes the throwaway stuff. The data says otherwise.
Can an AI Detector Catch a Fake Legal Citation Before It Gets You Sanctioned?
A law firm representing a major bank just got its brief struck for fake AI citations. Here's what a detector can and can't catch in a legal filing.
Why 99.9% of Scientists Who Use AI to Write Don't Disclose It
A 5.2-million-paper PNAS study found 70% of journals have AI writing policies, but just 0.1% of papers disclose using AI. Here's the gap, and the pushback.
Newspapers Are Now Running AI Detectors on Every Opinion Submission. The Numbers Don't Agree With Each Other.
Two audits ran the same detector on newspaper opinion pages. US papers: 16% AI-touched. Dutch papers: 42%. Here's why the gap is real.
PwC's AI-Written Report Invented an Entire Product. Four Governments "Used" a Framework That Doesn't Exist.
PwC, KPMG, EY, and Deloitte all published AI-written reports with fabricated citations. A detection score alone wouldn't have caught it. Here's the real fix.
89% of Biomedical Papers Now Show Signs of AI Writing. Here's the Number Editors Should Actually Trust.
A new corrected method finds 89% of Dec 2025 biomedical papers show AI-writing signs. Here's what editors should actually take from it.
Why an AI Detector Can Miss AI Text Without Anyone Trying to Fool It
Two 2026 studies (RAID benchmark, UF security team) found ordinary settings like repetition penalties and homoglyphs cut detector accuracy, no evasion needed.
Can AI Detectors Tell the Difference Between Writing With AI and Editing With AI?
New 2026 research on hybrid and humanised text shows most detectors miss AI-assisted writing that isn't purely AI or purely human.
How to Actually Evaluate an AI Detector's Accuracy Claims
A University of Chicago study and an FTC order both show why one accuracy number can't tell you if an AI detector actually works for you.
A "100% Human-Written, Never AI" Company Turned Out to Be Entirely AI. Here's How to Actually Check.
Research Gold promised "100% human, never AI" research help. Its staff, sales emails, and phone rep were all AI. Here's how to verify that claim yourself.
Why "AI-Polished" Writing Can Score 100% AI-Generated (It's Not a Bug)
Nature ran its own news sites through an AI detector. Human-written, AI-edited articles scored 100% AI. Here's the mechanism that causes it.
The Base-Rate Problem: Why a 99% Accurate AI Detector Can Still Be Wrong Most of the Time
A new peer-reviewed paper does the math schools and publishers keep skipping: when misuse is rare, even a near-perfect detector mostly flags innocent people.
Can an AI Detector Actually Catch an AI-Written Resume?
Hiring managers say they can spot an AI resume. The surveys behind that claim never test accuracy.
Peer Reviewers Are Quietly Using AI to Write Reviews. Detectors Are Catching It.
A cancer-research publisher's AI detector found reviewers using AI to write review reports far more than they admitted. Here's what the data shows.
Pew Says a Third of New Web Pages Are AI-Written. Here's What That Actually Means
Pew Research measured AI authorship across 490,000 webpages. Here's what the numbers show, and what a population stat can't tell you about one page.
Can You Appeal an AI Detection Accusation? What the UK's Ombudsman Just Ruled
The UK's OIA ruled on 4 AI-misconduct cases in 2025. 3 of 4 involved international students, and the pattern isn't detector error.
Does a GPTZero Flag Hold Up in Court? What the Yale Lawsuit Actually Shows
A 13-count federal lawsuit against Yale shows what really decides an AI-cheating case: not the detector score, but everything around it.
AI Detectors Barely Miss ChatGPT. They Miss AI Trained to Write Like a Specific Person.
AI detectors catch plain AI text almost every time. A new Epoch AI benchmark shows they miss up to 1 in 5 passages built to imitate one author's style.
Confirming AI Flags Doesn't Work: 5 Ways Schools Try Anyway (None of Them Prove Anything)
A peer-reviewed paper shows 5 ways schools try to confirm an AI flag are all confirmation bias, not real evidence.
Does NIH Actually Detect AI in Your Grant Application?
NIH runs an undisclosed AI detector on grants with real clawback power. NSF just asks for disclosure. A new PNAS study shows what each produces.
Substack's New AI Detector Is Already Flagging Real Writers. Here's the Actual Lesson.
Substack's new Pangram-powered detector is flagging real human writers. Here's why a strong detector still does that, and what a score actually means.
LinkedIn's New "Seems Like AI Slop" Button Is a Vote, Not a Detector
LinkedIn's new AI-slop flag button is a crowd vote, not a detector. Here's the accuracy gap real detectors close that a click never can.
New AI Transparency Laws Just Kicked In. They Don't Cover a Single Word of Text.
New AI transparency laws took effect Aug 2, 2026. An audit found most compliance detectors failed on edited files. Text isn't covered at all.
Why AI-Generated Job Applications Are Slowing Down Hiring (And What Actually Helps)
67% of HR leaders say AI-written applications have slowed hiring. Volume isn't the fix recruiters need — verification is.
Claude Now Watermarks AI Text. Here's Why You Still Need a Detector.
Anthropic's new Claude watermark signals AI processing, not authorship, and vanishes under paraphrase. Detection still fills the gap.
Teens Wrote a National AI-in-Schools Policy. It Doesn't Ban Detectors — It Makes Them Fairer.
98 students passed a K-12 AI policy that requires detectors plus mandatory human review, not a ban.
A 99% Accurate AI Detector Still Isn't Proof of Anything. Here's Why.
A novelist's book got pulled and a WSJ editor got accused. Both traced to one detector's score misused as proof.
Why Are Universities Turning Off AI Detectors? What the Backlash Actually Means
Yale, Vanderbilt and NYU limited AI-detector scores as sole evidence in 2026. What changed, and what to do if you get flagged.
Can Hiring Managers Actually Tell When Your Cover Letter Was Written by AI?
Most hiring managers say they can spot an AI cover letter. The research says the cover letter stopped working as a signal at all.
AI Detector False Positives: What the 2026 Evidence Actually Shows
The Authors Guild tested pre-2023 writing and found detectors flagging it as AI. Here's what causes false positives and how to respond.
How to Detect AI in Student Essays: A Step-by-Step Process
A detector score alone isn't proof. Here's the workflow: score, writing history, conversation, before any grade changes.
How to Prove You Didn't Use AI to Write Something
Version history is the strongest evidence you have. Here's how to document your writing process before anyone ever asks you to.
Why AI Detectors Are Biased Against STEM Writing and Code
A 280,000-sample study found AI detectors fail on code and unfairly flag disciplined STEM writing. Here's what the research actually found.
One in Three New Science Papers Now Reads as AI-Written. Here's What That Number Actually Means
A new 12,750-paper analysis finds a third of arXiv submissions now read as AI-written. Here's what that number means, and what it can't tell you.
Why Does My Essay Get Flagged as AI? What to Actually Do About It
A flag is a probability estimate, not a verdict. What actually triggers false positives, a real case, and steps to take before you panic.
AI Detection Score Meaning: What That Percentage Actually Tells You
A 62% AI score is not 62% of your text. It's a probability, not a fraction. Here's what detector scores actually measure, and how to read one.
AI Watermark Detection Explained: Why It Barely Works for Text (Yet)
OpenAI and Google's 2026 provenance push watermarks images, video, and audio at scale. Text watermarking exists too, but paraphrasing already breaks it.
AI Generated Text Examples: What the Fingerprints Actually Look Like
Real examples of AI text's statistical fingerprints, quoted from peer-reviewed studies with actual frequency ratios and accuracy numbers, not a vibe check.
ChatGPT Detector for Teachers: How to Actually Use One Without Getting It Wrong
A detection score is a starting point for a conversation, not a verdict. Here's how teachers use these tools well, and where it goes wrong.
Is AI Detection Accurate? What the 2026 Research Actually Shows
Two 2026 studies vs the 2023 baseline: false positives are down sharply, but hybrid text still defeats most detectors.
Does 'Humanizing' AI Text Actually Beat Detection? What the Research Shows
'Humanizer' tools promise undetectable AI writing. Peer-reviewed research tells a messier story about what really happens when you run it through a detector.
Can Turnitin Detect ChatGPT? What Turnitin's Own Data Actually Says
Yes, most of the time — but Turnitin's own numbers reveal a real gap between its accuracy claim and its false-positive rate.
Can AI Detectors Be Wrong? What the Research Actually Shows
Yes. Peer-reviewed research and real 2025 cases show who gets falsely flagged and how to use a detection score responsibly, not as a verdict.
Does Grammarly Count as AI? The Question Every Student and Teacher Keeps Getting Wrong
Grammarly bundles a grammar checker with generative AI writing help. Academic-integrity policy hasn't caught up, and it's costing students real grades.
7 Telltale Signs a Piece of Writing Was Generated by AI (and Why They're Not Enough on Their Own)
Repeated transitions, even sentence rhythm, hedge-everything phrasing: 7 real patterns in AI writing, and why spotting them by eye isn't proof.
How AI Detectors Actually Work: Perplexity, Burstiness, and the Statistics Behind the Score
AI detection isn't a magic trick, it's applied statistics. What perplexity and burstiness measure, and why no single signal tells the whole story.