Check out [this high-level article](https://prostasia.org/blog/csam-filtering-options-compared/) about tools for identifying CSAM.
Here are a few tools:
- [PhotoDNA](https://www.microsoft.com/en-us/photodna): Microsoft converts an image into a hash for **known** CSAM images. These images are illegal to have or host.
- They have a free API that people can get access to via approval process. It seems like the process is really meant for platforms and not researchers, so it is a bit annoying to deal with.
- This has been the industry standard for a while, however, it seems like it may be a bit outdated as we know that there are ways to trick it.
- [Thorn Safer](https://safer.io/): [Thorn](https://www.thorn.org/) is very well known company that specializing in combatting CSAM. "**Safer**" is one of their "CSAM Detection" products. I believe it relies on some sort of machine learning method but I don't know.
- Also have an API.
- I emailed with one of the sales folks their and their tools are not free.
- See [this page](https://aws.amazon.com/marketplace/pp/prodview-dfwekn4bx4ake) for pricing.
- [Facebook's PDQ](https://about.fb.com/news/2019/08/open-source-photo-video-matching/): This is Facebooks in-house algorithm. Also works for videos.
- Open source but not API-based and, thus, more complicated to implement.
---
#### Related
[[tools]] [[data_tools]]