#1 for true-positive rate, NIST Child Online Safety, ages 13-16See the evidence
Examine what AI can create and conceal.Models
43
Published for images, video, audio and text, each with its scope and limits.
From $99 a month, monthly minimum commitmentDetector24 pricing
Detector24 brings specialist detection models to images, video, audio and text, with distinct signals for different kinds of evidence.
AvailableEach model's scope and limits are published on its own page.
Talk to the Detector24 team
Detect signs of AI-generated images, image forgery and deepfakes, alongside a broader range of moderation and analysis capabilities.
Trinity focuses on identifying fully AI-generated images. Other specialist models address edited imagery and manipulated media, giving applications distinct signals for different kinds of evidence.
A photo can be real and edited, or generated from nothing. Treating those as separate questions gives a reviewer a clearer answer than one blended score.
Available. Catalogue entries are published by Detector24; each model's scope and limits are documented on its own page.
How it works
A photo can be real and edited, or generated from nothing. Detector24 asks each question with its own model, so a reviewer sees which one was answered.
01In
02Detector24
Generated or not
Trinity looks for images created entirely by generative models.
Edited or not
Forgery models look for the part of a genuine photo that was changed.
One signal each
Every model returns its own result instead of one blended score.
03Out
Result
One edited region, marked on the photo
Two of Detector24's jobs, step by step: telling generated and edited media apart, and moderating a continuous stream of content.
Trinity looks for images created entirely by generative models, without relying on watermarks or metadata.
Models for AI-edited image forgery, where part of a genuine photo has been changed, and for image deepfakes.
Video deepfake detection and face liveness, voice deepfake detection and AI-generated music detection.
Document tampering detection, document liveness, document classification and OCR for document images.
Fraud text, AI-generated text and requests for personal information, along with sentiment and fake news detection.
Image, video and text models for nudity and adult content, violence, weapons and other categories a platform needs to manage.
Published by the portfolio entry
43Published catalogue entries
27 image models, 6 video, 3 audio and 7 text, from AI-generated image detection and document tampering detection to voice deepfake and fraud text detection.
Detector24 model catalogue. Catalogue as published on 15 October 2026.
View the model catalogueTrinity is also a DecisionNode booster, so a decision can take the likelihood that an image was generated into account.
Image-authenticity signals sit alongside document forensics in ZOREAL Verify and agentic fraud reasoning when Bynn examines the evidence behind a claim or an application.
ZOREAL Verify also examines images for generation and manipulation, as part of its document checks. Detector24 remains its own portfolio entry, with specialist models for every kind of media.
These capabilities were offered on bynn.com. Detector24's specialist models carry them forward.
AvailableHuman trust for digital interactions: identity verification and document forensics in ZOREAL Verify, verified meetings, sign-in and signing.
Explore ZOREAL
AvailableAI for age assurance, built on Bynn's facial age-estimation research.
Explore Agemin
AvailableA decision model and inference service that returns decisions connected systems can act on.
Explore DecisionNode
ResearchA mission executive and autonomy engine for autonomous systems, demonstrated in simulation.
Explore Zora
Tell us what you need to check, from generated images to fraud text, and we will point you to the right models.