Skip to content

Company

GPTZero

GPTZero is a company that makes AI-generated text detection tools, widely used by educators and publishers to flag machine-written content.

Current stories

buildConfirmed5 publishers

Anthropic's Watermark Answers The Wrong Question, And Its Own Docs Say So

The Claude watermark survives light editing and dissolves under heavy editing, and its absence proves nothing. Policies that treat it as a verdict are built on a binary that does not exist.

Perspective Coverage

5 publishers
Builder
Builder 41%
Operator
Operator 47%
Investor
Investor 12%

Reality

Evidence55
Adoption45
Hype gap+5
Incentives35
Confidence60
securityOne report1 publisher

Reviewers policing OpenAI's training contractors are barred from using AI detection tools

404 Media found contractors who read real ChatGPT conversations have been fired for handing in model-generated work, and the internal documents show the sanctioned check on them is a human judging the overall pattern.

Publishers:404media.co

Reality

Evidence64
Adoption58
Hype gap+12
Incentives62
Confidence66
productOne report1 publisher

Substack hands its readers a button that scans your post for AI text

Fast Company reports that Pangram beats rival detectors in independent testing, without naming an error rate, and Substack, NewsGuard and an exam platform are already running it on writing whose authors never opted in.

Publishers:fastcompany.com

Reality

Evidence42
Adoption58
Hype gap+32
Incentives66
Confidence48
buildOne report1 publisher

StealthGPT claims 89% Pangram bypass rate in own benchmark, separately reports 18% factual-consistency failure rate

Super exists because Pangram V4 stopped working against StealthGPT's old rewriter, and the launch report puts the evasion number next to a factual-consistency score of 82 percent on the same 100 samples.

Publishers:runtimewire.com

Reality

Evidence32
Adoption27
Hype gap+41
Incentives86
Confidence57
productOne report1 publisher

Hachette cancelled a novel after a startup's model scored it 78 percent AI

Pangram has 24 employees and a percentage that Substack now shows to readers, so publishers and prize juries are making irreversible calls on a number checked mainly in early independent tests, by a company that has since shipped a new model.

Publishers:wired.com

Reality

Evidence42
Adoption55
Hype gap+38
Incentives72
Confidence48
scienceOne report1 publisher

AI text detectors are good enough to deploy. The appeals process is what nobody has written.

Cancer-journal editors are already catching undisclosed LLM use in peer review with commercial software. The unresolved question is what happens to a flagged researcher who says no.

Publishers:nature.com

Reality

Evidence64
Adoption68
Hype gap+22
Incentives74
Confidence55
investOne report1 publisher

63% of Amazon's religious titles look AI-written, and the only gate is a checkbox

Originality.ai flagged 1,272 of 2,034 books. The audit's soft spots and Amazon's honour-system disclosure point the same way: nobody is verifying who wrote what.

Publishers:decrypt.co

Reality

Evidence41
Adoption54
Hype gap+24
Incentives74
Confidence44