AI detectors by content type
AI detectors by content type is a view of one field, modalities, which records what each vendor says its detector accepts: text, code, image, voice, video or music. All 72 cards state at least one, so every card appears in the matrix below. As read on 8 November 2026 for the first sixteen cards and 9 November 2026 for the twenty-six added that day, 34 cards stated text (SynthID's as a watermark check with no public checker, Illuminarty's as a proof of concept in the vendor's words), 43 after the nine text-only cards of the ninth wave on 10 November 2026 and 51 after the eight text cards of the tenth wave later the same day, 20 state image, 13 state video, 11 state voice, 4 state code and 4 state music at the ninth-wave count, and the tenth wave's four non-text cards (ElevenLabs Classifier for voice, FakeCatcher for video, Deepfake-O-Meter for image, video and voice, Ircam Amplify for music) make it 21 image, 15 video, 13 voice and 5 music, and the eleventh wave later the same day (seven text cards, GPT Radar, Grammica, IvyPanda, SEO.ai, AI Writing Check, RADAR and Ghostbuster, and two voice cards, Hiya Deepfake Detector and McAfee Deepfake Detector) makes it 58 text and 15 voice; eight cards, Codequiry, Was It AI, Sightengine, Deepware, Sensity, Codeleaks, Pindrop Pulse and Resemble Detect, state no text at all. The fifth wave on 9 November 2026 added the index's first two voice-first detectors, Pindrop Pulse and Resemble Detect, a second code-only card, Codeleaks, and a third image detector with a study record, Illuminarty. The third and fourth waves of cards on 9 November 2026 brought in the index's first video-only detector, Deepware, and two more cards that state music, AI or Not and Sightengine, beside Hive. This page is one of three slices under AI detector features; the API slice and the LMS integrations slice filter the same cards on other fields, the code view goes deeper on the three code cards, and you can compare them side by side on the full board.
- Field 1Text
- Field 2Code
- Field 3Image
- Field 4Voice
- Field 5Video
- Field 6Music
The modalities field takes the vendor's own list from the product page, docs or launch post each card cites. A modality is marked only when the vendor names it as something the detector scores; a marketing sentence about "all content" does not count, and a modality the vendor offers only through an API is marked with that note. Winston AI's "handwriting (OCR)" is a text input read from an image, so the matrix records it under text and image. The field rules are in the methodology, and the independently tested ranking shows that every study in the index scored text and nothing else.
Which field decides the coverage matrix?
Values as each vendor states them, read 8 November 2026 for the first sixteen rows and 9 November 2026 for the rest. "Proof of concept" is Illuminarty's own wording for its text detector. "API" means the vendor offers that modality through its API only, as stated on the card. The text and image question, added 9 November 2026, reads the fifteen cards whose field lists both text and image and prints their text and image study counts side by side.
Why each card is in this view
- AIDI-D-0007QuillBot
text and image (quillbot.com, read 8 November 2026).
- AIDI-D-0008Grammarly
text (grammarly.com, read 8 November 2026).
- AIDI-D-0009Scribbr
text (scribbr.com, read 8 November 2026).
- AIDI-D-0005Winston AI
text, image and handwriting (OCR); uploads in .docx, .png and .jpg (gowinston.ai/pricing, read 8 November 2026).
- AIDI-D-0006Pangram
text, image and video; image JPG, PNG, WebP up to 30 MB, video MP4, MOV, WebM up to 500MB (pangram.com, read 8 November 2026).
- AIDI-D-0001GPTZero
text and image (AI Vision, specs not published) (gptzero.me, read 8 November 2026).
- AIDI-D-0003Copyleaks
text, code, image and video (copyleaks.com, read 8 November 2026).
- AIDI-D-0012Phrasly
text and image (phrasly.ai, read 8 November 2026).
- AIDI-D-0010Sapling
text (sapling.ai, read 8 November 2026).
- AIDI-D-0013JustDone
text; the FAQ mentions images but no image product exists on the site (justdone.com, read 8 November 2026).
- AIDI-D-0004Originality.ai
text and code (5.0 launch post, read 8 November 2026).
- AIDI-D-0011ZeroGPT
text, image and video (zerogpt.com, read 8 November 2026).
- AIDI-D-0014Proofademic
text, with image and voice through the API (proofademic.ai, read 8 November 2026).
- AIDI-D-0015Walter Writes
text, with image and voice through the API (walterwrites.ai/api); images up to 10 MB, audio up to 150 MB and 10 minutes (pricing), both read 8 November 2026.
- AIDI-D-0016Hive (now Deploy)
text, image, video, voice (audio) and music, the only card to state all five of the non-code modalities; audio is scored per 10-second chunk and music per 3-second chunk (docs, file formats, read 8 November 2026).
- AIDI-D-0002Turnitin
text; the FAQ states that lists, headers, references, tables, formulas and code are excluded from scoring (Turnitin FAQ, read 8 November 2026).
- AIDI-D-0017GLTR
text; the open-source demo colours each word by its rank under GPT-2 117M (gltr.io, read 9 November 2026).
- AIDI-D-0018AIDetectPlus
text, pasted or uploaded as .docx, .pdf, .txt or .html (aidetectplus.com/ai-detector, read 9 November 2026).
- AIDI-D-0019GPT-2 Output Detector
text; a RoBERTa classifier fine-tuned on GPT-2 outputs (model card, read 9 November 2026).
- AIDI-D-0020Codequiry
code only, in 65+ programming languages (codequiry.com, read 9 November 2026).
- AIDI-D-0021WasItAI
image only, up to 8MB and 10000x10000px (wasitai.com, read 9 November 2026).
- AIDI-D-0022Reality Defender
video, image, voice (audio) and text files; video up to 250 MB and 30 minutes, audio up to 20MB, images up to 50MB, text up to 900KB (SDK docs, read 9 November 2026).
- AIDI-D-0023Writer
text, pasted or by URL, on the archived page of a tool sunset on 22 December 2025 (web.archive.org, read 9 November 2026).
- AIDI-D-0024Undetectable AI
text, up to 10,000 words per web check (undetectable.ai, read 9 November 2026).
- AIDI-D-0025AI or Not
image, video, voice (audio), music and text, one file per web check and separate API endpoints per type (aiornot.com, read 9 November 2026).
- AIDI-D-0026Grammarly Authorship
text, as a record of how it was typed, pasted or generated inside Google Docs, Word or Grammarly docs; no probability score (grammarly.com/authorship, read 9 November 2026).
- AIDI-D-0027Turnitin Clarity
text written inside the Clarity writing space, with AI writing results only through the Originality add-on (guides.turnitin.com, read 9 November 2026).
- AIDI-D-0028SynthID
image, video and audio through the Detector portal and Gemini; text only as an open-source watermark check with no public checker, marked Watermark in the matrix (synthid.com, read 9 November 2026).
- AIDI-D-0029Smodin
text, pasted or uploaded as .txt, .doc, .docx or .pdf (smodin.io, read 9 November 2026).
- AIDI-D-0030BrandWell
text, by paste, URL or file on the archived page and by paste in the Moxby Mod (web.archive.org, read 9 November 2026).
- AIDI-D-0031Surfer
text, English only (surfer.ai, read 9 November 2026).
- AIDI-D-0032Compilatio
text, plus handwriting through OCR on handwritten documents in Magister (compilatio.net, read 9 November 2026). The card's modalities field lists text and handwriting (OCR) and no image, so the image column is blank since 9 November 2026: OCR of a handwritten page reads the words so the text detector can score them and is not a check of whether an image was generated. Winston AI lists image and handwriting as two entries, so its image cell stays.
- AIDI-D-0033Quetext
text, up to 1,000 words per free scan (quetext.com, read 9 November 2026).
- AIDI-D-0034TruthScan
text, image, PDF documents, video (deepfake) and voice (audio), the images and PDF pages counted as results on the plans (truthscan.com, read 9 November 2026).
- AIDI-D-0035DetectGPT
text, scored by perturbation with a source model and T5 from the MIT-licensed code (arXiv 2301.11305, read 9 November 2026).
- AIDI-D-0036Sightengine
image, video, deepfake, voice (audio speech) and music through one API; no text (sightengine.com, read 9 November 2026).
- AIDI-D-0037Deepware
video only, face manipulation; the FAQ says the scanner does not detect voice, images or text (deepware.ai, read 9 November 2026).
- AIDI-D-0038Sensity
video, image, voice (audio) and file forensics; no text (sensity.ai, read 9 November 2026).
- AIDI-D-0039Codeleaks
source code only, 11 programming languages named; the AI code verdict was withdrawn from the Copyleaks API on 16 December 2025 per the release notes (copyleaks.com/codeleaks, read 9 November 2026).
- AIDI-D-0040Illuminarty
image, with a text classifier the vendor calls a proof of concept; the one card scored on images by a study beside Hive and AI or Not, at 72.65% accuracy and a 67.40% false positive rate in Ha and Passananti 2024 (illuminarty.ai, read 9 November 2026).
- AIDI-D-0041Pindrop Pulse
voice (audio) on live calls and uploaded recordings, and video, synthetic avatars and face swaps in Pulse for Meetings (pindrop.com, read 9 November 2026).
- AIDI-D-0042Resemble Detect
voice (audio), image, video and music, with Suno and Udio listed as covered music generators and 17 counted upload formats (resemble.ai/products/detect, read 9 November 2026).
- AIDI-D-0043Decopy
text; the same suite sells a separate AI Image Detector and AI Code Detector not read for the card (decopy.ai/ai-detector, read 9 November 2026).
- AIDI-D-0044Content Detector AI
text, on the archived page of 11 May 2026 (web.archive.org, read 9 November 2026).
- AIDI-D-0045GPTKit
text; image files are accepted as uploads of text, not as image detection (gptkit.ai, read 9 November 2026).
- AIDI-D-0046PlagiarismCheck
text, English and Spanish (plagiarismcheck.org/ai-detector, read 9 November 2026).
- AIDI-D-0047Isgen
text; the suite's AI Image Detector and AI Video Detector are separate tools not read for the card (isgen.ai, read 9 November 2026).
- AIDI-D-0048AI Detector Pro
text; the same company's Codespy code detector is a separate product not read (aidetector.pro/pages/faq, read 10 November 2026).
- AIDI-D-0049Fast-DetectGPT
text, a zero-shot method on token probabilities (arXiv 2310.05130, read 10 November 2026).
- AIDI-D-0050Paperpal
text, scholarly writing (paperpal.com/tools/ai-detector, read 10 November 2026).
- AIDI-D-0051Detecting-AI.com
text, with TXT, PDF and DOCX read in the browser (detecting-ai.com, read 10 November 2026).
- AIDI-D-0053Binoculars
text, a research method scored through two Falcon-7B models (github.com/ahans30/Binoculars, read 10 November 2026).
- AIDI-D-0054Monica
text, through three third-party engines the page names (monica.im/tools/ai-content-detector, read 10 November 2026).
- AIDI-D-0055Trinka
text, from a Word or PDF upload (trinka.ai/features/ai-content-detector, read 10 November 2026).
- AIDI-D-0056StealthWriter
text, pasted (stealthwriter.ai/ai-detector, read 10 November 2026).
- AIDI-D-0057Wordtune
text, on the archived page of a withdrawn tool (web.archive.org, 30 March 2025, read 10 November 2026).
- AIDI-D-0058Check For AI
text and files, on the archived page of a tool sunset in 2023 (web.archive.org, 27 May 2023, read 10 November 2026).
- AIDI-D-0059Writefull
text, on the archived page of a discontinued tool (web.archive.org, 31 December 2024, read 10 November 2026).
- AIDI-D-0060Passed.ai
text, pasted or uploaded, English only, through Originality.ai's engine; the service is closing (web.archive.org, 7 June 2026, read 10 November 2026).
- AIDI-D-0062ElevenLabs Classifier
voice, for ElevenLabs-generated speech only; audio from other providers is stated as not detected (elevenlabs.io/ai-speech-classifier, read 10 November 2026).
- AIDI-D-0063FakeCatcher
video, portrait videos read through blood-flow signals; no public access route (intel.com newsroom, read 10 November 2026).
- AIDI-D-0064Deepfake-O-Meter
image, video and voice, through 41 hosted research detectors with a login; the audio detectors are speech models, so music is not claimed (zinc.cse.buffalo.edu/ubmdfl/deep-o-meter, read 10 November 2026).
- AIDI-D-0065Ircam Amplify
music, AI-generated tracks from eleven named generators (ircamamplify.com, read 10 November 2026).
- AIDI-D-0066GPT Radar
text, a withdrawn paste tool (web.archive.org, 12 July 2026, read 10 November 2026).
- AIDI-D-0067Grammica
text, a withdrawn free tool (web.archive.org, 31 July 2025, read 10 November 2026).
- AIDI-D-0068IvyPanda
text, a colour-coded essay checker (ivypanda.com, read 10 November 2026).
- AIDI-D-0069SEO.ai
text, in 50+ stated languages (seo.ai, read 10 November 2026).
- AIDI-D-0070AI Writing Check
text, a withdrawn nonprofit tool (web.archive.org, 17 July 2023, read 10 November 2026).
- AIDI-D-0071RADAR
text, a research method with a Hugging Face Space (github.com/IBM/RADAR, read 10 November 2026).
- AIDI-D-0072Ghostbuster
text, a research method with a demo page (github.com/vivek3141/ghostbuster, read 10 November 2026).
- AIDI-D-0073Hiya Deepfake Detector
voice, the voice in a video or audio playing in a Chrome tab (chromewebstore.google.com, read 10 November 2026).
- AIDI-D-0074McAfee Deepfake Detector
voice, AI-generated audio in a video playing in the browser, scored on the PC's NPU (mcafee.com, read 10 November 2026).
What the evidence says by content type
Every study record in the index scored text. Walters 2023 scored 126 essays (publisher PDF), Weber-Wulff 2023 scored 54 documents (arXiv), Perkins 2024 scored 114 samples (arXiv), Elkhatat 2023 scored 35 paragraphs (Springer PDF), Liang 2023 scored 179 essays (arXiv), and RAID 2024 scored a corpus of generated prose (arXiv), all read 8 November 2026 and listed in the studies index. On that text, Copyleaks and Turnitin classified all 126 Walters 2023 documents correctly, Originality.ai (model 1.4) 98% with 0 incorrect, and GPTZero (Pro) 81% with a 2% false positive rate (Walters 2023, publisher PDF, read 8 November 2026). In Weber-Wulff 2023, Winston AI scored 36 of 54 documents with 0 false accusations in 18 human-written documents and ZeroGPT 32 of 54, also with 0 (Weber-Wulff 2023, arXiv, read 8 November 2026).
No study record evaluates code, image, voice, video or music detection, and the only non-text figure any card carries is the vendor's: Originality.ai reports no false positives among 150 human-written code samples in its 5.0 launch post (launch post, read 8 November 2026). A text result says nothing about the same vendor's image or voice model. No detector score should be the only evidence in a misconduct or hiring decision, and for a non-text score there is no published study behind the number at all.
Nine cards list voice and video in one product (Hive, AI or Not, Sightengine, Reality Defender, SynthID, TruthScan, Sensity, Pindrop Pulse, Resemble Detect) and none of the nine has an indexed record on either medium, as the voice and video question, added 9 November 2026, sets out.
Two cells in the matrix are read by newer pages: the handwriting statements of Winston AI and Compilatio are the first table of the handwriting and OCR view, and the five music cells (Ircam Amplify, Hive, Resemble Detect, Sightengine, AI or Not) are the music detectors question, which records that no study in the index scores AI-generated music; the scanned PDF question reads the PDF and OCR statements across the text column, all read 10 November 2026.
Where do the non-text modalities go for depth?
The index catalogues text detectors that also state another modality; a reader whose job is a single non-text modality has specialist tools to compare. For image, the nineteen image cards above (twenty before Compilatio's handwriting OCR entry was moved out of the image column on 9 November 2026) range from a listed word to published file specs, with Was It AI the one image-only card, and a dedicated comparison lives at aiimagedetectorapi.com. For voice, four cards state it, two of them through an API only, and aivoicedetector.com covers voice detection on its own. For video, five cards state it, Reality Defender being the one built around deepfake video, and aivideodetector.io is the specialist reference. For music, Hive is the single card, and aimusicdetector.ai compares tools built for that modality. The two code cards are compared on the code view.
Sources
- https://quillbot.com/ai-content-detector (read 8 November 2026)
- https://www.grammarly.com/ai-detector (read 8 November 2026)
- https://www.scribbr.com/ai-detector/ (read 8 November 2026)
- https://gowinston.ai/pricing/ (read 8 November 2026)
- https://www.pangram.com/ (read 8 November 2026)
- https://gptzero.me/ (read 8 November 2026)
- https://copyleaks.com/ai-content-detector (read 8 November 2026)
- https://phrasly.ai/ai-detector (read 8 November 2026)
- https://sapling.ai/ai-content-detector (read 8 November 2026)
- https://justdone.com/ai-detector (read 8 November 2026)
- https://originality.ai/blog/introducing-originality-5-0-and-max (read 8 November 2026)
- https://www.zerogpt.com/ (read 8 November 2026)
- https://proofademic.ai/?utm_source=aidetectorindex.com&utm_medium=owned&utm_campaign=index_view_row (read 8 November 2026)
- https://walterwrites.ai/api/?utm_source=aidetectorindex.com&utm_medium=owned&utm_campaign=index_view_row (read 8 November 2026)
- https://walterwrites.ai/pricing/?utm_source=aidetectorindex.com&utm_medium=owned&utm_campaign=index_view_row (read 8 November 2026)
- https://docs.thehive.ai/docs/ai-generated-content-detection (read 8 November 2026)
- https://docs.thehive.ai/docs/supported-file-formats (read 8 November 2026)
- https://guides.turnitin.com/hc/en-us/articles/28477544839821-Turnitin-s-AI-writing-detection-capabilities-FAQs (read 8 November 2026)
- https://www.degruyterbrill.com/document/doi/10.1515/opis-2022-0158/pdf (read 8 November 2026)
- https://arxiv.org/pdf/2306.15666 (read 8 November 2026)
- https://arxiv.org/pdf/2403.19148 (read 8 November 2026)
- https://link.springer.com/content/pdf/10.1007/s40979-023-00140-5.pdf (read 8 November 2026)
- https://arxiv.org/pdf/2304.02819v3 (read 8 November 2026)
- https://arxiv.org/pdf/2405.07940 (read 8 November 2026)
- http://gltr.io/ (read 9 November 2026)
- https://aidetectplus.com/ai-detector (read 9 November 2026)
- https://huggingface.co/openai-community/roberta-base-openai-detector (read 9 November 2026)
- https://codequiry.com/detect-ai-written-code (read 9 November 2026)
- https://wasitai.com/ (read 9 November 2026)
- https://docs.realitydefender.com/sdks/quickstart (read 9 November 2026)
Revision log: 2026-11: view created; 2026-11-09: six cards added to the matrix; 2026-11-09: sixteen cards from the third and fourth waves added, 38 rows; 2026-11-09: four wave 5 cards added, 42 rows, two voice-first; the file types ranking counts the upload formats behind each claim2026-11-09: Compilatio's image cell cleared to follow the card's modalities field (text and handwriting OCR, no image), with the OCR note; the text and image question linked (seventh wave)2026-11-09: eighth-wave pages linked2026-11-10: nine ninth-wave cards added, all text only; 43 text cards2026-11-10: twelve tenth-wave cards added: eight text only, ElevenLabs Classifier (voice), FakeCatcher (video), Deepfake-O-Meter (image, video, voice) and Ircam Amplify (music); 63 rows2026-11-10: nine eleventh-wave cards added: seven text only, Hiya Deepfake Detector and McAfee Deepfake Detector (voice); 72 rows2026-11-10: twelfth-wave pages linked
FAQ: AI detectors by content type
Which AI detectors handle more than text?
Eleven of the 22 cards state a second modality as read on 8 and 9 November 2026: QuillBot, Winston AI, Pangram, GPTZero, Copyleaks, Phrasly, ZeroGPT, Proofademic, Walter Writes, Hive and Reality Defender. Grammarly, Scribbr, Sapling, JustDone, Originality.ai (text and code), Turnitin, GLTR, AIDetectPlus and the GPT-2 Output Detector state no image, voice, video or music input; Codequiry states code alone and Was It AI image alone.
Which AI detectors detect images?
Twelve cards: QuillBot, Winston AI, Pangram, GPTZero, Copyleaks, Phrasly, ZeroGPT and Hive in the product, Proofademic and Walter Writes through an API, all as stated on 8 November 2026, and Was It AI and Reality Defender, added 9 November 2026. Pangram and Walter Writes publish file specs; GPTZero states that AI Vision specs are not published.
Which AI detectors detect voice or audio?
Four cards: Hive in the product, scored per 10-second chunk, and Proofademic and Walter Writes through an API, as stated on 8 November 2026, plus Reality Defender, whose SDK takes audio up to 20MB, read 9 November 2026. Walter Writes accepts audio up to 150 MB and 10 minutes. For a voice-only comparison, aivoicedetector.com is the specialist reference.
Which AI detectors detect video or music?
Video: Pangram, Copyleaks, ZeroGPT, Hive and Reality Defender, five cards. Music: Hive alone, scored per 3-second chunk with named attribution classes, all as stated on 8 November 2026. No study in the index has evaluated either modality.
Does a detector's text accuracy apply to its image or code model?
No. Each modality is a separate model, and every study indexed before 9 November 2026 scored text. A card with 7 study records on 8 November 2026, such as GPTZero, carried 7 records about its text detector and none about AI Vision, and its 11 records on 9 November 2026 are still all text. Read the figure as a statement about the modality the study tested. The one record on another modality, Ha and Passananti 2024 on Hive's image detector, was added 9 November 2026, and the ranking by content type counts records per modality for the fourteen cards that state a non-text type.
). 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)
