Dr. Neal Krawetz, one of the leading experts in the area of computer forensics research, digital photo analysis, and related topics, has penned a blog post in which he takes apart Apple’s recent announcement and the technology behind it.
He actually has a lot of experience with the very problem Apple is trying to deal with, since he is the creator of FotoForensics, and files CSAM reports to the National Center for Missing and Exploited Children (NCMEC) every day. In fact, he files more reports than Apple, and knows all the ins and outs of all the technologies involved – including reverse-engineering Microsoft’s PhotoDNA, the perceptual hash algorithm NCMEC and Apple are using.
The reason he had to reverse-engineer PhotoDNA is that NCMEC refused to countersign the NDA’s they wanted Krawetz to sign, eventually not responding to his requests altogether. Krawetz is one of the more prolific reporters of CSAM material (number 40 out of 168 in total in 2020). According to him, PhotoDNA is not as sophisticated as Apple’s and Microsoft’s documentation and claims make it out to be.
Perhaps there is a reason that they don’t want really technical people looking at PhotoDNA. Microsoft says that the “PhotoDNA hash is not reversible”. That’s not true. PhotoDNA hashes can be projected into a 26×26 grayscale image that is only a little blurry. 26×26 is larger than most desktop icons; it’s enough detail to recognize people and objects. Reversing a PhotoDNA hash is no more complicated than solving a 26×26 Sudoku puzzle; a task well-suited for computers.
The other major component of Apple’s system, an AI perceptual hash called a NeuralHash, is problematic too. The experts Apple cites have zero background in privacy or law, and while Apple’s whitepaper is “overly technical”, it “doesn’t give enough information for someone to confirm the implementation”.
Furthermore, Krawetz “calls bullshit” on Apple’s claim that there is a 1 in 1 trillion error rate. After a detailed analysis of the numbers involved, he concludes:
What is the real error rate? We don’t know. Apple doesn’t seem to know. And since they don’t know, they appear to have just thrown out a really big number. As far as I can tell, Apple’s claim of “1 in 1 trillion” is a baseless estimate. In this regard, Apple has provided misleading support for their algorithm and misleading accuracy rates.
Krawetz also takes aim at the step where Apple manually reviews possible CP material by sending them from the device in question to Apple itself. After discussing this with his attorney, he concludes:
The laws related to CSAM are very explicit. 18 U.S. Code § 2252 states that knowingly transferring CSAM material is a felony. (The only exception, in 2258A, is when it is reported to NCMEC.) In this case, Apple has a very strong reason to believe they are transferring CSAM material, and they are sending it to Apple — not NCMEC.
It does not matter that Apple will then check it and forward it to NCMEC. 18 U.S.C. § 2258A is specific: the data can only be sent to NCMEC. (With 2258A, it is illegal for a service provider to turn over CP photos to the police or the FBI; you can only send it to NCMEC. Then NCMEC will contact the police or FBI.) What Apple has detailed is the intentional distribution (to Apple), collection (at Apple), and access (viewing at Apple) of material that they strongly have reason to believe is CSAM. As it was explained to me by my attorney, that is a felony.
This whole thing looks, feels, and smells like a terribly designed system that is not only prone to errors, but also easily exploitable by people and governments with bad intentions. It also seems to be highly illegal, making one wonder why Apple were to put this out in the first place. Krawetz hints at why Apple is building this system earlier in this article:
Apple’s devices rename pictures in a way that is very distinct. (Filename ballistics spots it really well.) Based on the number of reports that I’ve submitted to NCMEC, where the image appears to have touched Apple’s devices or services, I think that Apple has a very large CP/CSAM problem.
I think this might be the real reason Apple is building this system.