Cycling through all six. Tap any point to stop.

Measured on six things

Trusted AI detector for ChatGPT, Claude, and Gemini.

Security

HTTPS enforced and SPF present, but no HSTS, CSP, frame protection, or Referrer-Policy — 3/12 on the header checklist. No DMARC. No SOC 2, ISO, or bug-bounty page found.

Screenshots

1 / 1

AI or Not

About this app

AI or Not is a free AI detector that instantly analyzes text to determine if it was written by a human or generated by AI models like ChatGPT, Claude, and Gemini. It provides an overall AI-probability score with a confidence level, and for supported languages, a sentence-by-sentence breakdown highlighting which parts are likely AI-generated. The tool is designed to protect real writers by minimizing false positives and treating scores as signals to weigh with context. It supports up to 500,000 characters per check for free, and a free account unlocks higher daily limits and check history. Additional detectors for images, video, voice, and music are available.

Key features

1Free AI text detection with no signup required, up to 500,000 characters per check.
2Overall AI-probability score with a confidence level on every result.
3Sentence-by-sentence highlighting of likely AI-generated passages for supported languages.
4Free account unlocks higher daily limits and check history.
5Additional detectors for images, video, voice, and music available.

Target customer profile

Drawn from the product itself, not from a survey.

Demographic

Students and academics

Pain points

Fear of unintentional plagiarism or AI-generated content in submissions

Primary needs

Verify originality and ensure work is human-written before submission

Demographic

Content creators and marketers

Pain points

Maintaining authenticity and credibility with audience

Primary needs

Ensure content reads naturally and avoids AI detection flags

Demographic

Educators and institutions

Pain points

Detecting AI-generated assignments and maintaining academic integrity

Primary needs

Reliable detection tool with detailed analysis for fair evaluation

Scout Review

Written by AI from measured evidence, scored out of 100.

AI or Not is a free, no-signup AI text detector that returns an overall AI-probability score with a confidence level and, for supported languages, sentence-by-sentence highlighting. It supports up to 500,000 characters per check for free, and a free account raises daily limits and adds check history. The site is clean and modern, the example results panel is well-executed, and the desktop experience is fast. Mobile performance is the weak spot at 73/100 with a 4.5-second LCP. Security headers are thin — 3/12 on the probe — and there's no SOC 2, ISO, or bug-bounty signal. Accessibility scores 96/100 with one contrast failure. The core idea is sound and the target customers are clearly named, but the category is crowded with GPTZero, Originality.ai, and Copyleaks, and the differentiators here are features, not moats. A solid, competent entry that needs to harden its security posture and sharpen its positioning to stand out.

Measure by measure

Usability

Paste-and-detect with no signup is the right call. FAQ covers the obvious questions. Free account adds history and higher limits. No visible pricing page, so paid path is unclear.

Design

Clean, modern dark hero with a paste box and clear CTA. Example Results block with red-flagged sentences sells the concept well. Card-based feature sections are tidy but generic SaaS template territory.

Speed

Desktop Lighthouse 96/100 is strong. Mobile 73/100 with LCP 4.5s and FCP 4.4s drags it down — mobile users wait nearly five seconds for first paint. Detection itself claims seconds.

Security

HTTPS enforced and SPF present, but no HSTS, CSP, frame protection, or Referrer-Policy — 3/12 on the header checklist. No DMARC. No SOC 2, ISO, or bug-bounty page found.

Accessibility

Lighthouse a11y 96/100 with one failing audit: insufficient color contrast. Multi-language detection claimed. No WCAG statement or a11y page found, so keyboard/screen-reader depth is unverified.

Growth

Real, growing problem with a specific wedge — sentence-level flags, confidence scores, and a 500k-character free tier. But GPTZero, Originality.ai, and Copyleaks already own education and SEO mindshare.

Improvement areas

Mobile performance: Lighthouse 73/100 with a 4.5s LCP — the biggest measurable drag.
Security headers: 3/12 on the probe, no HSTS, CSP, or DMARC, and no security page.
No visible pricing page or paid-tier comparison, leaving the paid path opaque.
Accessibility: one contrast failure and no WCAG statement or a11y page.

Design & Usability

The site is a competent, modern SaaS landing page: dark hero, paste box, clear 'Detect AI' button, and an Example Results panel that shows an 87% score with red-flagged sentences and per-sentence percentages. That example does more selling than the rest of the page combined. Feature cards for Research Papers, Reports, Articles, and School Assignments map cleanly to the stated target customers. Usability-wise, the no-signup free check is the right call for a tool people want to try before trusting, and the FAQ covers model coverage, languages, and the honest caveat that detectors can be wrong. The gaps: no visible pricing page in the fetched content, so the paid path is opaque, and the free-account benefits (higher limits, history) are described in prose rather than a comparison table. Navigation is shallow and the page is long, but nothing is broken. Against GPTZero and Copyleaks, the design is on par or slightly ahead; usability is on par.

Speed & Security

Speed is split down the middle. Desktop Lighthouse performance lands at 96/100, which is genuinely fast. Mobile lands at 73/100 with LCP 4.5s and FCP 4.4s — nearly five seconds before first paint on a phone, which is the device most students and casual checkers will use. CLS is fine at 0.037 and TBT is a healthy 89ms, so the problem is asset delivery, not JavaScript execution. The detection itself claims 'seconds,' and there's no reason to doubt that. Security is the weak flank. HTTPS is enforced and SPF exists, but the HTTP probe found no HSTS, no Content-Security-Policy, no frame protection, no X-Content-Type-Options, no Referrer-Policy, and no Permissions-Policy — 3/12 on the checklist. DMARC is also missing. There is no SOC 2, ISO 27001, or bug-bounty page surfaced anywhere on the site. For a tool whose entire value proposition is 'paste your unpublished essay here,' that's a thin posture. None of this is unusual for a half-year-old indie tool, but it's also not nothing.

Accessibility & Growth

Accessibility is the strongest measured dimension: Lighthouse scores 96/100, with a single failing audit for insufficient color contrast. That's a real, fixable flaw, not a systemic one. The site claims detection across English, Chinese, and 'most other languages,' which is a meaningful accessibility-adjacent feature for non-English users. What's missing is any accessibility statement, WCAG conformance claim, or evidence of screen-reader and keyboard testing beyond what the automated audit caught. On growth: the problem is real and getting bigger — AI-generated text is everywhere and institutions need to verify it. The wedge is specific: sentence-level highlighting, a confidence level on every score, and a 500,000-character free limit that undercuts most competitors' free tiers. The problem is the neighborhood. GPTZero owns education mindshare, Originality.ai owns SEO and content marketing, and Copyleaks has institutional sales. AI or Not's differentiators are features the incumbents can copy in a sprint. The domain is six months old, so there's no traction story yet — this is a positioning play, and the positioning is decent but not defensible.

Conclusion

If you need a quick, free sanity check on whether a passage reads as AI-generated, AI or Not does the job with no signup and a generous character limit. The sentence-level breakdown is the most useful part — a single percentage is easy to argue with, but 'this specific sentence reads as AI' is actionable. The caveats are real: mobile load times are sluggish, the security headers are minimal for a tool that handles unpublished writing, and the competitive field is stacked. Treat the score as a signal, not a verdict, exactly as the site itself advises. It's worth bookmarking as a free second opinion. It is not yet worth paying for over the incumbents until the security posture and the paid offering are clearer.

Suggestions for improvement

Fix the mobile LCP — 4.5s first paint is the single biggest measurable drag on the product.
Add HSTS, CSP, and Referrer-Policy headers; publish a security page even if it's just a plain statement.
Publish a real pricing page with a feature comparison table instead of describing limits in prose.
Add an accessibility statement and fix the contrast failure flagged by Lighthouse.

How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

	AI or Not	GPTZero	Originality.ai	Copyleaks
Free tier character limit per check	500,000 characters per check, no signup required	Limited free scans; paid plans for higher volume	Credit-based; no large free tier	Free trial credits; paid subscription for volume
Sentence-level AI highlighting	Yes, for supported languages, with per-sentence percentages	Yes, sentence-level highlighting is a core feature	Yes, sentence-level AI detection scoring	Yes, sentence-level AI content detection
Confidence level on results	Yes, confidence level shown with every score	Probability score; confidence framing varies	Probability score per scan	Probability score with AI content percentage
Multi-language detection	English, Chinese, and 'most other languages' claimed	Multiple languages supported	Primarily English-focused	30+ languages supported
Additional media detectors (image/video/audio)	Image, video, voice, and music detectors available	Text-focused; limited media detection	Text and AI image detection	Text, code, and some media detection
Plagiarism checking	Not offered; AI detection only	Not a core offering	Yes, plagiarism checking included	Yes, plagiarism detection is a core feature