What they are making available is sufficient to 'cheat' the system in the sense that if you have an image you are concerned might match in some database you can modify it until the 'perceptual hash', which you can compute on your own, changes. The novel changed image is then unlikely to be a match in the database.
You don't have to have a copy of the database to be fairly confident that your modifications have made a target image non-matching. You would have to have the database in order to gather evidence that the matching was beginning to be used for unlawful, rights violating purposes, such as collecting targets for genocide.
I think it's a safe assumption that this sort of system is only effective against idiots-- which isn't an argument against it: lots of effective anti-crime measures mostly work against idiots. Adding functionality which destroys accountability which at most improves the system against non-targets, however, doesn't seem like an acceptable trade-off.
You don't have to have a copy of the database to be fairly confident that your modifications have made a target image non-matching. You would have to have the database in order to gather evidence that the matching was beginning to be used for unlawful, rights violating purposes, such as collecting targets for genocide.
I think it's a safe assumption that this sort of system is only effective against idiots-- which isn't an argument against it: lots of effective anti-crime measures mostly work against idiots. Adding functionality which destroys accountability which at most improves the system against non-targets, however, doesn't seem like an acceptable trade-off.