AI or Not AI detector
Vendor AIorNot, Inc. Official site aiornot.com How this card is sourced
Updated · AI Detector Index editors
AI or Not flagged 24.47% of 280 human artworks as AI-generated in the 2024 Ha and Passananti study, which tested the tool under its earlier name Optic (arXiv 2402.03214, Table 1, read 9 November 2026). The vendor is AIorNot, Inc. of California, and the product sells one credit pool across image, deepfake, voice, music, video and text checks, priced per image and per second of media (aiornot.com/pricing, read 9 November 2026). The card joins the detector index beside the other cards built on media rather than prose, and how this card is sourced explains why its false positive figure comes from an image test and not a text one.
What AI or Not detects
| Field | Value | Source |
|---|---|---|
| Content types detected | Image, video, voice, music, text | aiornot.com/pricing |
| Generators covered | Nano Banana, Midjourney, DALL-E / GPT Image, Stable Diffusion, Flux, Grok / Aurora, Ideogram, Adobe Firefly, Recraft, Seedream, Qwen, StyleGAN and GAN models (image coverage last updated July 2026), ChatGPT, Claude and Meta LLaMa (text) | aiornot.com/ai-image-detector |
| Input limits | Web tool: one file per check, image, video or audio in MP3, MP4, PNG, JPG, WEBP or GIF; text pasted or uploaded as DOC, DOCX or TXT. API: images up to 50MB in jpg, jpeg, png, webp, heic, heif or tiff (10MB in the docs playground); video up to 200MB in 17 formats, with the synchronous endpoint analysing a portion of videos longer than 30 seconds; audio in common formats such as MP3 and WAV, with responses of up to two minutes on long files; text from 250 characters (about 64 words) to 500,000 characters per request. | docs.aiornot.com/api-reference/reports-by... |
| Free tier | Free plan: $5 in free credits, 1M words of AI text detection free every month, 20 AI image checks, API key included; no daily limit stated for image checks on a free account | aiornot.com/pricing |
| API | Yes, docs.aiornot.com | docs.aiornot.com |
| Integrations | REST API (api.aiornot.com/v2), MCP server (Claude, Cursor, Codex), AI or Not: Live Voice for Zoom, Reverse image search report | aiornot.com/ai-image-detector |
| Data retention | API documentation: Privacy first: uploaded media are processed and deleted immediately after inference. Free text tool: No ads. No data stored. Privacy Policy: We retain Personal Data about you for as long as you have an open account with us or as otherwise necessary to provide you with our Services; Live Voice for Zoom meeting audio is held only in volatile memory and discarded after inference, with operational logs retained for up to 31 days. | docs.aiornot.com |
| Launch year | 2023 | aiornot.com/press |
| Ownership | AIorNot, Inc. (terms of use; privacy policy signed AI or Not Inc.), governed by California law with small claims venue in Alameda County, California; site copyright 2023 to 2026. The vendor's press page links June 2023 coverage of the tool as the Optic web app, and Ha and Passananti 2024 tested it under the name Optic. Founder named on the press page: Anatoly Kvitnitsky. | aiornot.com/terms-of-service |
| Independent studies | 1 record | Studies index |
The home page describes the service as an "AI Detection API For All Content Types" that will "Detect AI-generated images, text, videos, and deepfakes" (aiornot.com, read 9 November 2026). The pricing page adds voice and music to that list under the Pro plan, which "unlocks text, image, deepfake, voice, music, and video detection" (aiornot.com/pricing, read 9 November 2026). The API documentation describes separate reports by modality: image, audio voice, audio music, video and text, with the image endpoint returning ai_generated, deepfake, nsfw and quality reports by default and a reverse_search report on request (docs.aiornot.com, read 9 November 2026).
Image coverage is named. The image page lists Nano Banana, Midjourney, DALL-E / GPT Image, Stable Diffusion, Flux, Grok / Aurora, Ideogram, Adobe Firefly, Recraft, Seedream, Qwen and "StyleGAN & GAN models", with "Generator coverage last updated: July 2026", and states that detection "is based on the pixel content of the image, not metadata, watermarks or provenance signals such as EXIF or C2PA content credentials" (aiornot.com/ai-image-detector, read 9 November 2026). The text tool "specializes in detecting content from ChatGPT, Claude and Meta's LLaMa" and returns "a confidence score" rather than a verdict (aiornot.com/ai-detection, read 9 November 2026). No language list or count is published for text, so the languages field is empty.
Input limits come from the API reference, read 9 November 2026: images up to 50MB in jpg, jpeg, png, webp, heic, heif or tiff; video up to 200MB in 17 formats, where the synchronous endpoint "will only analyze a portion of the content for longer duration videos" and the vendor recommends 30-second segments; audio in common formats such as MP3 and WAV with responses of up to two minutes on long files; and text from 250 characters, "approximately 64 words", to 500,000 characters per request (text endpoint, video endpoint). The web tool takes one file in MP3, MP4, PNG, JPG, WEBP or GIF, or text as DOC, DOCX or TXT. Integrations are the REST API, an MCP server for "Claude, Cursor or Codex", and a Live Voice app for Zoom described in the privacy policy (privacy policy, read 9 November 2026). The sibling media cards are Hive, tested in the same 2024 study, and Reality Defender.
Limits and plans
| Plan | Price | Limits | Checked on |
|---|---|---|---|
| Pro (monthly) | $5/month | $10 in credits included every month, credits roll over monthly; unlocks text, image, deepfake, voice, music and video detection; credit equivalents 2M words, 500 image checks or 1,000 seconds of video; pay-as-you-go credits. | |
| Pro (yearly) | $50/year | $100 in credits instantly, valid for the full term; credit equivalents 20M words, 5,000 image checks or 10,000 seconds of video. | |
| Business (monthly) | $100/month | $100 in credits every month; 25% less than Pro on image, audio and video checks; credit equivalents 20M words, 6,600 image checks or 13,300 seconds of video; reseller rights up to $5,000/month; no contract. | |
| Business (yearly) | $1,000/year | $1,000 in credits instantly, valid for the full term; credit equivalents 200M words, 66,000 image checks or 133,000 seconds of video. | |
| Pay-as-you-go credit rates | Pro: $0.02 per image, $0.01 per second of video, $0.005 per second of audio, $5 per million words of text; Business: $0.015 per image and 25% less on audio and video, text and reverse image search undiscounted | Credits work across every model; monthly plans roll unused credits over while the plan is active. | |
| Enterprise | Custom; on-prem deployment, forward deployed AI researcher for custom models, model calibration, full reseller rights, private instances, dedicated support. |
Every check is billed in credits, and the pricing page states the rates in its FAQ: "Pro rates are $0.02 per image, $0.01 per second of video, $0.005 per second of audio, and $5 per million words of text", with Business paying "25% less on image, audio and video checks" so that "$0.02 per image becomes $0.015", while text and reverse image search "are not discounted" (aiornot.com/pricing, read 9 November 2026). The free plan gives "$5 in free credits", "1M words of AI text detection free every month", "20 AI image checks" and an API key. Figures below are as displayed, read 9 November 2026; the yearly toggle is labelled "2 months free".
Retention is stated shortest in the developer documentation: "Privacy first: uploaded media are processed and deleted immediately after inference" (docs.aiornot.com, read 9 November 2026), and the free text tool page says "No ads. No data stored." (aiornot.com/ai-detection, read 9 November 2026). The card's retention rank field takes the documentation clause as a stated zero. The privacy policy governs account data instead: "We retain Personal Data about you for as long as you have an open account with us or as otherwise necessary to provide you with our Services", and Zoom meeting audio "is held only in volatile memory for real-time analysis and is discarded after inference" with operational logs kept "for up to 31 days" (privacy policy, read 9 November 2026). Ownership is AIorNot, Inc. under the terms of use, with California law and small claims venue in Alameda County (terms of use, read 9 November 2026).
What the vendor claims
Is it AI? Detect AI-generated images, text, videos, and deepfakes via API with 98.9% accuracy.
The home page states an accuracy rate of 98.9%: "Is it AI? Detect AI-generated images, text, videos, and deepfakes via API with 98.9% accuracy", beside counters of "7M+ Checks Run" and "500K+ Users" (aiornot.com, read 9 November 2026). The image page repeats the sentence "Accuracy in identifying images of all image types" under a counter that did not render a number to this reader. No false positive rate, sample size, test date or dataset accompanies the 98.9% on any vendor page.
The vendor's own caveats are operational. The text tool returns "a confidence score giving you the likeliness of whether the text was written by a person or artificial intelligence" (aiornot.com/ai-detection, read 9 November 2026); text checks need at least 250 characters; and the video endpoint analyses only a portion of a long file (docs.aiornot.com, read 9 November 2026). On robustness the image page says: "Our models are updated frequently for new adversarial attacks and major generative releases, so your confidence (score) holds up against the latest generators" (aiornot.com/ai-image-detector, read 9 November 2026). That is the vendor's sentence; the study below measured the same question.
What independent studies found
- AIDI-S-0017Ha and Passananti 20242024
Optic (AI or Not): 90.67% accuracy, 24.47% false positive rate and 1.15% false negative rate on 280 human artworks and 350 AI-generated images from five generators (Table 1), with 4.29% of decisions in the dropped 40% to 60% band; detection of AI images fell from 98.85% to 62.62% under medium Glaze and 58.00% under high (Table 3).
One record in the studies index tested AI or Not, under the name Optic. Ha and Passananti 2024 ran 280 human artworks and 350 AI-generated images from five generators through the tool and recorded 90.67% accuracy, a 24.47% false positive rate and a 1.15% false negative rate, after dropping the 4.29% of decisions that fell in the 40% to 60% "not sure" band (arXiv 2402.03214, Table 1, read 9 November 2026). The authors write that Optic's low false negative rate "comes at the expense of a high 24.47% FPR", against Hive's 0% on the same artworks.
The same paper measured the adversarial question the vendor raises. Under Glaze, the protection tool for artists, the share of AI images Optic detected fell from 98.85% to 62.62% at medium intensity and 58.00% at high intensity (arXiv 2402.03214, Table 3, read 9 November 2026). The vendor's press page links June 2023 coverage that names the web app Optic and its founder Andrey Doronichev, which is the vendor's own confirmation that the Optic result belongs on this card (aiornot.com/press, read 9 November 2026). The study covers images only; no indexed record has scored the tool's text, audio or video detection.
The gap between the vendor's 98.9% claim (aiornot.com, read 9 November 2026) and the 90.67% accuracy in Ha and Passananti 2024 (arXiv 2402.03214, Table 1) is 8.2 points on the one study available, and that study predates the July 2026 generator coverage the vendor now lists. A reader should treat the 2024 numbers as a measurement of the tool as it stood in early 2024.
Where AI or Not appears in the index
Rankings
- AIDI-R-01AI detectors ranked by independent testing Two records added in the fifth wave on 9 November 2026, Pegoraro 2023 and Fleckenstein 2024, changed nine counts and none: Pegoraro 2023 tested ZeroGPT, GPTZero, Copyleaks, Originality.ai, Writer, Content at Scale (BrandWell), DetectGPT, the GPT-2 Output Detector and GLTR, lifting GPTZero to 12, the GPT-2 Output Detector to 10, Copyleaks to 7 and Originality.ai and ZeroGPT to 6; Fleckenstein 2024 tested 289 teachers and no software, so it counts for no card ([Fleckenstein 2024 record](/studies/fleckenstein-2024-teachers/)). Illuminarty, one of four cards added in the same wave, joined with its one image record from Ha and Passananti 2024, so 18 of 72 cards now qualify.
- AIDI-R-03AI detector free tiers, compared
- AIDI-R-05AI detectors ranked by maximum input per check
- AIDI-R-06AI detectors ranked by stated data retention
- AIDI-R-07AI image, voice, video and code detectors ranked by independent studies
- AIDI-R-09AI detector false negative rates, ranked
- AIDI-R-10AI detector accuracy claims vs published results, ranked by the gap
- AIDI-R-11AI detectors ranked by named integration surfaces
- AIDI-R-12AI detectors ranked by the upload formats the vendor states
- AIDI-R-13AI detectors ranked by minimum text length per check
- AIDI-R-14AI detector APIs ranked by published price per 1,000 words
- AIDI-R-17AI detectors ranked by the generator models the vendor names
- AIDI-R-20AI detectors ranked by launch year, with every change of hands and withdrawal the cards record
Fits these use cases, questions and comparisons
- AIDI-U-09AI detectors for law firms and courts: an index fit check
- AIDI-U-11AI text detectors for newsrooms and fact desks
- AIDI-U-13AI detectors for government and public sector buyers
- AIDI-U-14AI detectors for self-publishing authors: the cards that take a manuscript and say what happens to it
- AIDI-UP-03Studies digest, Q4 2026: the nine evaluations added since launch, with one headline figure each
- AIDI-UP-09Index update, Q3 2027: twenty-one cards added, two fields added, two rows held, three log lines
AI or Not ranks in seven lists, read 9 November 2026 in the data file. AI detectors ranked by independent studies counts its 1 record; false positive rates sorts on the 24.47% of 280 artworks; false negative rates on the 1.15% of 350 AI images; free tiers on the 1,000,000 free text words per month; data retention on the stated zero days; studies by modality on its one image record; and the claim versus evidence gap on the 8.2 points between 98.9% and 90.67%.
AI or Not also appears on AI detectors by content type, where text, image, voice, video and music are columns, and on detectors with an API. It is absent from the language ranking, which needs a published count, and from the robustness ranking, which measures paraphrased text rather than Glazed images. Every detector card on the register shows the same fourteen fields, so the per-second media pricing here sits beside the per-word pricing of the text tools.
AI or Not is one of fifteen cards on the text and image question, added 9 November 2026, one of ten on the free API question on its Free plan of 1M words a month with an API key, and seventh on the generators covered ranking on 15 named generators.
Questions about AI or Not AI detector
How much does AI or Not cost?
Pro is $5/month or $50/year and Business is $100/month or $1,000/year, with credits spent at $0.02 per image, $0.01 per second of video, $0.005 per second of audio and $5 per million words on Pro, read 9 November 2026 on aiornot.com/pricing.
Is AI or Not free to use?
Yes, within limits: the Free plan carries $5 in credits, 1M words of text detection a month and 20 image checks, with an API key, per the pricing page read 9 November 2026.
What was AI or Not's false positive rate in an independent test?
24.47% of 280 human artworks were flagged as AI in Ha and Passananti 2024, which tested the tool as Optic; its false negative rate was 1.15% of 350 AI images and its accuracy 90.67% (Table 1).
Does AI or Not keep the files you upload?
The API documentation says uploaded media "are processed and deleted immediately after inference", and the text tool page says "No data stored"; account data is kept while the account is open, per the privacy policy read 9 November 2026.
Which AI image generators does AI or Not detect?
Nano Banana, Midjourney, DALL-E / GPT Image, Stable Diffusion, Flux, Grok / Aurora, Ideogram, Adobe Firefly, Recraft, Seedream, Qwen and GAN models, with coverage last updated July 2026, per the image page read 9 November 2026.
Sources and revisions
Sources
- aiornot.comRead 9 November 2026
- aiornot.com/pricingRead 9 November 2026
- aiornot.com/ai-image-detectorRead 9 November 2026
- aiornot.com/ai-detectionRead 9 November 2026
- aiornot.com/pressRead 9 November 2026
- aiornot.com/privacy-policyRead 9 November 2026
- aiornot.com/terms-of-serviceRead 9 November 2026
- docs.aiornot.comRead 9 November 2026
- docs.aiornot.com/api-reference/reports-by...Read 9 November 2026
- docs.aiornot.com/api-reference/reports-by...Read 9 November 2026
- docs.aiornot.com/api-reference/reports-by...Read 9 November 2026
- arxiv.org/abs/2402.03214Read 9 November 2026
Revision log
- Card created2026-11-09: text and image question, free api question and generators covered ranking linked (seventh wave)
). Illuminarty, one of four cards added in the same wave, joined with its one image record from Ha and Passananti 2024, so 18 of 72 cards now qualify.: register plate](/img/independently-tested-480.webp)
