Freemium·AI-content detectors
Detects AI-generated text, source code, and plagiarism, and (since June 2026) AI-generated video and audio via a dedicated AI Video Detector that performs simultaneous, independent scans of a video file's visual and audio tracks to identify the exact moments AI-generated content appears. Targets academic institutions, enterprises, and SEO professionals, with the video detector aimed additionally at KYC, refund-fraud, and insurance-claims use cases.
Approach & detailsHide details
ApproachMulti-layered algorithm evaluating perplexity, burstiness, and linguistic fingerprinting to measure sentence structure irregularity and predictability against known AI patterns.Distinct fromOne of the few major detectors that extends its capabilities beyond text to explicitly scan programming code and integrate deeply with LMS platforms.Source: Copyleaks Official Site
Claims over 99% accuracy. Generally maintains a low false-positive rate on raw AI output, but effectiveness declines significantly with manual humanizing. Launched an AI Video Detector on June 2, 2026 (enterprise-grade, frame-level scanning of video and audio tracks independently, flagging fully synthetic content, AI speech paired with real footage, and hybrid AI-altered splices), materially extending Copyleaks beyond its original text/code/plagiarism scope; predates this lens's 2026-07-27 window but not previously reflected in this entry.
Paid·AI-content detectors
Enterprise platform (GetReal Protect) that verifies the authenticity of people in files and real-time digital communications, aimed at defending against deepfake fraud and identity manipulation; co-founded by digital-forensics researcher Hany Farid.
Approach & detailsHide details
ApproachCombines forensic content analysis for deepfake and manipulated-media detection across video, voice, and images with continuous identity authentication and adaptive policy enforcement, for both stored files and live calls.Distinct fromUnlike single-modality audio tools (Pindrop Pulse, Resemble Detect) or file-only forensic scanners, GetReal combines continuous real-time identity verification with deepfake detection across voice, video, and images spanning both live calls and stored files.Source: GetReal Security
Enterprise/paid product, demo-gated; company site lists backers Forgepoint Capital, Ballistic Ventures, Evolution Equity, Cisco, IQT, and Capital One. No independently verified accuracy figures found; none are claimed on the homepage.
Freemium·AI-content detectors
Detects AI-generated text and checks writing quality, aimed at educators and editors; offers writing-origin reports and learning-management-system integrations.
Approach & detailsHide details
ApproachAI-text detection using statistical signals (perplexity and burstiness) plus trained classifiers to label text as human, AI, or mixed across major models (ChatGPT, Claude, Gemini, Llama).Distinct fromUnlike a watermark-remover (which defeats detection) or a provenance standard (which cryptographically signs origin), GPTZero is a classifier that scores whether existing text is AI-written.Source: GPTZero (gptzero.me)
Acquired by Superhuman (formerly Grammarly) on June 23, 2026; GPTZero's AI and hallucination detectors are being integrated into Superhuman Go as an authenticity layer. Co-founders Edward Tian and Alex Cui joined Superhuman. Vendor claims 99% accuracy and 17 million users; self-reported and fabrication-prone, not independently verified.
Paid·AI-content detectors
Detects AI-generated text, images, deepfakes, video, and audio. Targets enterprise developers and platforms for automated moderation workflows.
Approach & detailsHide details
ApproachAPI-first infrastructure that scans for algorithmic artifacts left by generative AI. Uses probability scores based on massive, diverse datasets of synthetic and organic content.Distinct fromExcels in multimodal detection (images, deepfakes, audio) and operates at the infrastructure API level rather than as a consumer web app.Source: Hive Moderation
Vendor claims high accuracy (often cited around 98% for images), but text detection remains subject to standard AI detector limitations and false positives.
Paid·AI-content detectors
Detects AI-generated text from major LLMs and plagiarism. Targets publishers, agencies, and high-volume content teams.
Approach & detailsHide details
ApproachAnalyzes text for statistical predictability and patterns using machine learning models trained on human and AI text. Introduces an AI Allowance feature for custom tolerance thresholds.Distinct fromBuilt for screening content at scale for publishers rather than purely academic settings. Features unique AI Allowance thresholding.Source: Originality.ai Official Site
Claims 99%+ accuracy on latest LLMs, but independent reviews note it can struggle with heavily edited or humanized text. Acknowledges false positives can occur.
Freemium·AI-content detectors
AI-text detection product (web dashboard, browser extension, Canvas/LMS integration, API) built for teachers, publishers, and trust-and-safety teams; became Substack's built-in AI-detection provider for posts over 100 words starting July 21, 2026.
Approach & detailsHide details
ApproachSentence-level linguistic and semantic analysis (not perplexity/burstiness) trained to catch AI text including lightly edited or humanized content; reports both a probability the text is AI-generated and an estimate of how much of the document is AI.Distinct fromUnlike GPTZero/Originality.ai/Copyleaks (perplexity-and-burstiness classifiers marketed mainly to educators or publishers as standalone checkers), Pangram markets itself on sentence-level explainability and a low false-positive rate, and is now the detection layer embedded directly into a major publishing platform's own AI-disclosure feature rather than a separate tool users visit.Source: Pangram (pangram.com)
Vendor-claimed 99.98% accuracy and third-party-reported ~1-in-10,000 false-positive rate are vendor/review-sourced, not independently reverified here. Free plan: 4 checks/day; paid subscription unlocks more credits and plagiarism detection. Substack integration announced 2026-07-21 via Substack's own product update, cited as evidence of adoption. Update (2026-07-29): Pangram raised $9M (led by Menlo Ventures, with Haystack, ScOp, Script Capital, and Cadenza participating) and launched Pangram 4, a next-generation text-detection model vendor-claimed at over 99% accuracy including improved detection of AI-humanizer output, plus Pangram Image, an AI-generated-image detector currently in research preview ahead of a wider release; this supersedes the prior 'text-only, no image detection' note.
Paid·AI-content detectors
Detects AI-generated voice clones and audio deepfakes, targeting contact centers, banks, and enterprise communications.
Approach & detailsHide details
ApproachLiveness detection and acoustic pattern analysis to identify mechanical signatures of synthetic voice generation, such as spectral distortions and frequency anomalies.Distinct fromAnalyzes audio physics and biological speech markers instead of evaluating text syntax or semantic predictability.Source: Pindrop Pulse
Vendor claims up to 99 percent accuracy on known models and 90 percent on zero-day attacks (unverified).
Paid·AI-content detectors
Detects AI-generated academic writing. Targets schools, universities, and educators directly integrated into their existing grading workflows.
Approach & detailsHide details
ApproachProbabilistic analysis engine measuring vocabulary distribution, sentence structure uniformity, and information flow to assign a probability percentage of AI involvement.Distinct fromEmbedded directly into the ubiquitous Turnitin plagiarism ecosystem, meaning students are automatically screened upon assignment submission without extra steps for teachers.Source: Turnitin AI Innovation
Accuracy is highly contested. Claims under 1% false positive rate for heavy AI usage, but real-world tests report higher false flags, particularly for non-native English speakers.
Freemium·AI-content detectors
Free-to-use, Iceland-based AI content detector site (founded 2023) offering dedicated tools for text, image, and, as of the August 11, 2026 launch, video content.
Approach & detailsHide details
ApproachStatistical AI-text classifier (perplexity/burstiness-family approach comparable to GPTZero) built as separate, purpose-built detection engines per content modality rather than one unified multimodal model; as of August 11, 2026 those engines cover text, images, and (new) video, evaluating visual frames, temporal patterns, compression signals, and artifacts associated with synthetic media across MP4, MOV, AVI, MKV, and WEBM.Distinct fromUnlike GPTZero (a separate company it is frequently confused with, launched about 15 days earlier in 2023) or AI or Not (a single classifier spanning four modalities from one company, Optic), ZeroGPT ships separate, dedicated detection engines per modality and only added video coverage in August 2026, three years after its text-only launch.Source: ZeroGPT Launches AI Video Detector, Expanding Detection Across Text, Images and Video (press release)
Quote attributed to Rawad Baroud, CEO and co-founder. ZeroGPT is an unfunded, Iceland-based company founded in 2023; vendor claims accuracy over 98%, but independent 2026 reviews (fast.io, aicheckr.io) report real-world accuracy in roughly the 60-85% range with false-positive rates up to ~33% depending on content type; not independently reverified here. Not previously tracked in this catalogue despite being a long-established GPTZero rival; the new video detector (Aug 11, 2026, within this lens's window) is the trigger for adding it now that it spans all three major content modalities.
Freemium·AI-content detectors
Free web detector that identifies AI-generated text, images, music, and video across the major generative models, made by Optic.
Approach & detailsHide details
ApproachMultimodal classifier that flags AI-generated content across text (ChatGPT, Claude), image (Midjourney, 4o), music (Suno, Udio), and video (Veo, Kling).Distinct fromUnlike single-modality detectors (GPTZero for text, Reality Defender for media), AI or Not covers all four content modalities in one tool.Source: AI or Not (aiornot.com, by Optic)
One of the best-known multimodal AI detectors; made by Optic.
Free·AI-content detectors
A standalone deepfake-detection consumer app from cybersecurity vendor Bitdefender, available across 14 countries, aimed at everyday consumers checking whether a video circulating online (for example, on social media) is a malicious deepfake.
Approach & detailsHide details
ApproachConsumer mobile app where a user pastes a video link or uploads a file for a structured, multi-layered analysis that reports likelihood of manipulation, transcript-level spoken-content analysis, and markers of deceptive intent (fraud, impersonation, credential theft) rather than a bare authenticity score.Distinct fromUnlike enterprise-focused forensic detectors aimed at platforms or governments (Reality Defender, Sensity AI, GetReal Security) or Slop Or Not's on-device iOS/Mac app (which covers text and images but not video), RealCheck is a mainstream consumer security-vendor app specifically for checking video links or files for deceptive-intent deepfakes, backed by an established antivirus brand's existing consumer install base.Source: Bitdefender: RealCheck, Your Personal Deepfake Detector
Announced 2026-06-24 (Bitdefender and Business Wire press release), predating this run's 2026-08-05 window; flagged as an untracked gap-fill rather than a confirmed new-in-window launch. Pricing and monetization model not disclosed in available coverage. Not previously tracked in this catalogue.
Paid·AI-content detectors
Detects deepfake videos, voice clones, and synthetic images, targeting governments, financial institutions, and media platforms for fraud prevention.
Approach & detailsHide details
ApproachPatented multimodal analysis using specialized neural networks to detect technical artifacts and behavioral cues in audio, video, images, and text.Distinct fromProvides comprehensive cross-modality detection for visual and audio media, rather than being limited to text-based LLM outputs.Source: Reality Defender
Vendor claims robust real-time detection without relying on biometric data (unverified).
Paid·AI-content detectors
Detects deepfake audio and verifies content authenticity, targeting telecom providers, government agencies, and entertainment studios.
Approach & detailsHide details
ApproachDeep learning models for speaker verification and multimodal detection, including proprietary watermarking that embeds unnoticeable signals in AI audio.Distinct fromCombines active watermarking with passive acoustic analysis to verify voice origin, unlike text plagiarism checkers.Source: Resemble AI
Vendor emphasizes a consent-first approach and claims high accuracy across various synthetic modalities (unverified).
Paid·AI-content detectors
Detects visual threats, deepfake videos, and synthetic identities (KYC), targeting cybersecurity firms, law enforcement, and B2B workflows.
Approach & detailsHide details
ApproachMultilayered detection utilizing pixel-level analysis, file structure examination, and metadata forensics to identify manipulation.Distinct fromFocuses on visual threat intelligence and forensic media manipulation instead of probabilistic text prediction.Source: Sensity AI
Marketed as a high-stakes forensic investigation tool, vendor claims to map coordinated misinformation campaigns (unverified).
Freemium·AI-content detectors
Detects AI text from ChatGPT, Claude, Gemini, LLaMA, and paraphrasing tools. Targets educators, publishers, and content creators.
Approach & detailsHide details
ApproachClassifier model predicting AI authorship probability. Provides sentence-by-sentence prediction maps, readability scoring, and uses OCR for document and image scanning.Distinct fromStrong focus on OCR technology to read handwritten or scanned documents, making it highly versatile for offline or traditional classroom assignments.Source: Winston AI Official Site
Markets a 99.98% accuracy rate, though independent benchmarks estimate lower real-world accuracy with some false positives on complex human prose.
Open source·AI-content detectors
A privacy-first, in-page detector for AI-generated images and video frames, requiring a small local Python backend for offline inference; adjustable confidence thresholds, no accounts or telemetry.
Approach & detailsHide details
ApproachBrowser extension that runs local inference (an ensemble of two Hugging Face vision classifiers) directly on images and video frames as a webpage renders, drawing red border overlays and confidence badges on content flagged as AI-generated, with no upload to any server.Distinct fromUnlike ai-or-not or hive-moderation (web-upload or API services users visit to check one file at a time), this inspects every image already rendered on whatever page a user is browsing, in real time, entirely on-device.Source: AI Slop Detector (browser extension)
MIT license, 7 stars at time of check; very small/early project, requires the user to run a local Python backend, so real-world adoption is minimal. Vendor-cited 85-95% accuracy range is unverified. Exact project age/launch date could not be confirmed; not previously tracked in this catalogue, flagged as an untracked find rather than confirmed as a brand-new 2026-07-27-window launch.
Paid·AI-content detectors
File and business-workflow validation platform, launched July 8, 2026, that lets organizations automatically check submitted media before it drives decisions (claims, onboarding, etc.); integrates Google AI content-detection signals as part of a broader validation framework.
Approach & detailsHide details
ApproachPairs a generative-AI reasoning layer with a deterministic heuristic and business-rules engine to validate submitted photos, documents, audio, and video against workflow context, rather than only flagging manipulation in isolation.Distinct fromUnlike pure deepfake classifiers (Reality Defender, Sensity AI) that return a bare authenticity score on a file, DeepScan frames the problem as business-workflow validation, combining forensic detection with configurable business rules and contextual checks.Source: Attestiv (GlobeNewswire release)
Pricing not disclosed on attestiv.com (customized plans per client). Launched July 8, 2026, just before this catalogue's prior 2026-07-11 run date; included since not previously tracked and materially distinct in framing from existing detector entries.
Paid·AI-content detectors
Real-time, on-device deepfake detection for live video calls, targeting HR/recruiting teams and high-value executives against video-interview and call-impersonation fraud; video detection is live now, voice detection is on the roadmap.
Approach & detailsHide details
ApproachOn-device model that runs locally in the background of Zoom, Teams, and Meet calls (no cloud upload) to flag synthetic faces and faceswaps in real time; built with Qualcomm hardware-optimization support.Distinct fromUnlike cloud-based, post-hoc deepfake detectors (Reality Defender, Sensity AI) or audio-only liveness tools (Pindrop Pulse), Halo runs entirely on-device in the background of a live video call rather than analyzing an uploaded file after the fact.Source: Scam.ai
Access model unconfirmed: the site offers a 'Try Halo' / waitlist flow rather than published pricing, so 'paid' here is a provisional best guess pending confirmed pricing, not a sourced claim. Launched via Qualcomm partnership announced at Computex 2026 (~2026-06-23, per businesswire.com), before this catalogue's 2026-07-11 prior run date; included since not previously tracked.
Paid·AI-content detectors
Real-time streaming audio deepfake detection module within Hydaway Digital Corp.'s RealityChek platform, targeting phone calls, contact centers, broadcast, and voice-authenticated identity checks.
Approach & detailsHide details
ApproachContinuous forensic analysis of a live audio stream using models tuned to spectral patterns, prosodic markers (rhythm, pitch, stress, timing), compression artifacts, and generative-model signatures associated with current voice synthesis and cloning technologies.Distinct fromUnlike Pindrop Pulse or Resemble Detect (established acoustic/liveness detectors also targeting contact centers and banks), Hydaway is a newly announced real-time streaming-specific detector marketed as purpose-built for live audio streams rather than post-hoc file analysis.Source: Hydaway (RealityChek Streaming Audio Detection)
Confirmed verbatim; quote attributed to Karl Kottmeier, CEO of Hydaway. Announced July 9, 2026 by Hydaway Digital Corp., just before this lens's 2026-07-27 window; included as a gap-fill since not previously tracked and distinctly framed as real-time-streaming-first vs. Pindrop/Resemble's established file-based products. Pricing not disclosed on the company or product site; 'paid' is a provisional best guess for an enterprise B2B security product, not a sourced claim.
Paid·AI-content detectors
Detects deepfake videos in real-time, targeting news organizations, broadcasters, and social media platforms.
Approach & detailsHide details
ApproachPhotoplethysmography (PPG) analysis to track subtle, rhythm-based color changes in facial veins caused by blood flow, which AI fails to replicate accurately.Distinct fromExamines biological signals and blood flow in video pixels rather than text artifacts or simple image inconsistencies.Source: Intel Newsroom: Intel Introduces Real-Time Deepfake Detector
Vendor claims 96 percent accuracy and real-time inference in milliseconds (unverified).
Paid·AI-content detectors
GPU-accelerated, self-hostable video-authenticity detection microservice distributed to partners such as Wowza, whose Video Intelligence Framework embeds it to bring real-time synthetic-video detection into livestreaming workflows spanning more than 35,000 deployments in over 170 countries, across on-premises, edge, hybrid, and air-gapped environments.
Approach & detailsHide details
ApproachA NIM (NVIDIA Inference Microservice) within NVIDIA's AI for Media platform that analyzes video frame by frame on NVIDIA GPUs to produce a classifier score for whether a clip contains synthetic content, intended for newsroom and broadcaster review workflows before publication.Distinct fromUnlike Intel FakeCatcher (biological PPG blood-flow signal analysis) or Reality Defender/Sensity AI (cloud-hosted forensic classification services), NVIDIA's detector is a GPU-optimized, self-hostable inference microservice with published compression-robustness benchmarks (92% accuracy uncompressed, dropping to 82% at 50% compression, 22ms per 1080p frame on RTX hardware) built for embedding inside partners' existing video infrastructure like Wowza rather than as a standalone hosted product end users visit directly.Source: NVIDIA Blog: At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI
Announced at SIGGRAPH 2026 on July 20, 2026, roughly two weeks before this lens's 2026-08-05 window start; included as a gap-fill since it was not previously tracked in this catalogue and is materially distinct from existing detector entries, but flagged as pre-window for the assembler's attention. Pricing not disclosed on NVIDIA's site; 'paid' is a provisional best guess for an enterprise NIM microservice distributed via commercial partners like Wowza, not a sourced claim.
Paid·AI-content detectors
Real-time meeting-security product that joins enterprise video calls to flag deepfake voices, AI-generated impersonation, and PII leakage as they happen.
Approach & detailsHide details
ApproachThree proprietary detection engines running as a visible participant inside live Zoom, Google Meet, and Microsoft Teams calls, continuously scanning for AI-generated responses, deepfake voices, and exposed personally identifiable information, with participant-level threat scores and minute-by-minute analysis.Distinct fromUnlike Halo/Scam.ai (on-device, video-only face-swap detection with voice on the roadmap) or GetReal Security (a broader identity-verification and forensic platform), Meeting Guard is a visible in-call participant combining voice-deepfake detection with live PII-leak monitoring across all three major meeting platforms simultaneously.Source: Polygraf AI Meeting Guard
Confirmed verbatim; quote attributed to Yagub Rahimov, CEO of Polygraf AI. Launched July 15, 2026, just before this lens's 2026-07-27 window; included as a gap-fill since not previously tracked. Official product page (polygraf.ai) returned a 403 to automated fetch, so cited via Help Net Security's report of the launch. Pricing not disclosed in available coverage; 'paid' is a provisional best guess for an enterprise product, not a sourced claim.
Freemium·AI-content detectors
A consumer mobile/desktop app (iPhone and Mac, plus a free browser-based checker for English text and photos) that lets an individual check a photo or block of text for AI origin without uploading it anywhere.
Approach & detailsHide details
ApproachOn-device AI text and image detector for iPhone and Mac. Runs entirely locally using the Apple Neural Engine; for images it checks for known watermarks (Google and OpenAI SynthID) and C2PA/IPTC content-credential metadata alongside pixel-level artifact analysis, and separately scores pasted text for likely AI authorship.Distinct fromUnlike ai-or-not or hive-moderation (web-upload services processed on a vendor's server) or ai-slop-detector-webext (a browser extension that still requires a locally-run Python backend), Slop Or Not is a self-contained native iOS/macOS app doing on-device inference with no server round-trip and no separate backend to install.Source: Slop Or Not
App Store listing (apps.apple.com, id6744578596) and a Help Net Security product showcase dated 2026-08-13 both confirm current availability, but existing press coverage and the app's ID suggest the product itself predates this run's 2026-08-05 window; flagged as an untracked gap-fill rather than a confirmed brand-new in-window launch. Vendor-claimed accuracy figures (99.5% text, 94-100% image/watermark detection depending on method) are self-reported marketing claims from the vendor's own site, not independently verified, and are omitted from approach/owns accordingly.
