by Thorn
Child-safety detection from the nonprofit Thorn: hash matching against 153.9M+ verified CSAM hashes plus AI classifiers for new images, video and text.
Pick Safer if your platform hosts user images, video or messages and you need to find and report child sexual abuse material, both known and new. It is not a general moderation tool, so pair it with one. Thorn's customer page quotes teams at Slack and MediaLab using it.
Hash matching against a 153.9M+ verified CSAM database, paired with classifiers for material no one has hashed yet.
Its hash database aggregates 153.9M+ verified hashes from trusted sources (safer.io, read 22 Nov 2026), and Deploy, formerly Hive, says its own CSAM detection is powered by Thorn, so even rivals build on Safer rather than replace it.
Main trade-off
It is a single-purpose child-safety tool with no general moderation.
| Plan | Monthly price | What it includes |
|---|---|---|
| Safer API (Thorn-hosted, contact sales) | Custom | Safer Match hash matching via API, Safer Predict image, video and text classifiers |
| Safer Enterprise (self-hosted, contact sales) | Custom | Self-hosted Safer Match, Safer Predict CSAM classifier, Moderation interface with wellness features, Reporting service for NCMEC or RCMP accounts |
Thorn does not publish Safer prices. Access starts with a contact form, and Safer is offered both as a Thorn-hosted API and as the self-hosted Safer Enterprise package.
Safer by Thorn is a child sexual abuse material detection product from the nonprofit Thorn. Safer Match compares uploads against a database of verified CSAM hashes, Safer Predict classifies new images, video and conversations for exploitation and grooming, and a self-hosted package reports to NCMEC or the RCMP.
Safer Match hash matching
Checks images and video against Thorn's database of verified CSAM hashes with cryptographic and perceptual hashing, so altered copies of known material are still found.
Safer Predict image and video classifier
Flags potentially new CSAM with CSAM, pornography or benign labels and a precision level you set for escalation.
Safer Predict text classifier
Reads chat line by line and as a whole conversation, assigning a risk score per label so teams can target the riskiest accounts first.
Context labels
Adds nudity, sexual content and apparent maturity labels to suspected files so the most urgent cases are reviewed first.
Safer Enterprise
A self-hosted package with Match, Predict, a moderation interface with wellness features and a reporting service linked to NCMEC or RCMP.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| Quote | No | Web | 3.9 | |
| Quote | No | Web | 3.6 | |
| Free | Yes | Web | 3.8 | |
| Quote | No | Web | 3.9 |
Proprietary
Thorn keeps Safer's pricing private. Platforms fill in a contact form and Thorn tailors an offer to the deployment, hosted API or self-hosted.
No free plan or public demo is listed, which is normal for tools that handle child sexual abuse material. Access is arranged directly with Thorn.
Match finds known material by comparing hashes of uploads with hashes of verified CSAM. Predict uses machine learning to find material that has never been seen before, and also scores text conversations for grooming and sextortion.
Safer Enterprise includes a reporting service that connects to your organisation's NCMEC or RCMP account. You decide what to escalate and submit; Safer gives the tooling.
Not the same product, but related. Deploy, formerly Hive, says its CSAM Detection API is powered by Thorn's technology, while Safer is Thorn's own product with self-hosting, text classifiers and reporting built in.
Safer is the child sexual abuse and exploitation detection product of Thorn, the nonprofit that builds technology against online child abuse. It is sold to platforms that host user uploads or messages, and its customer page quotes trust and safety staff at Slack and MediaLab, along with OpenAI's child safety team on Thorn's work assessing AI models. Deploy, the company formerly called Hive, also builds its CSAM Detection API on Thorn's technology.
It works in two layers. Safer Match compares every uploaded image and video against a database of more than 153.9 million hashes of verified CSAM from trusted sources, using exact cryptographic hashes and Thorn's own perceptual hashing, which still matches an image that has been resized or slightly altered. Safer Predict then looks for material no one has hashed yet: an image and video classifier labels content as CSAM, pornography or benign, with extra context labels for nudity and apparent age, and a text classifier scores messages and whole conversations for CSAM trading, child access, sextortion, self-generated imagery requests and grooming.
It is for trust and safety teams that are legally and morally required to find and report this material: file-sharing and cloud storage services, messaging and workplace chat, social apps, dating platforms and AI image generators. The Safer Enterprise package runs self-hosted inside the customer's own infrastructure, with a review interface that includes moderator wellness features and a reporting service that connects to a platform's NCMEC or RCMP account.
Safer does one job. It does not moderate hate, violence, spam or adult content in general, so a platform still needs a general moderation layer such as Deploy Content Safety, Sightengine or Checkstep, or a free classifier such as the OpenAI Moderation API, whose maker offers its own CSAM image classifier to approved organisations. Hokai lists it with the other moderation tools in AI governance tools.
Product facts on this page (pricing, plans, integrations, compliance) are read from Thorn's own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.