Unfortunately, you probably need to add "and true" to that list, which makes the "non-deletable" requirement less clear-cut.

There is no algorithm for truth, so in practice you need a way of allowing groups to form consensus about the validity of claims relating to pieces of data in the system. A claim might be "The image with hash ABC123 is a deepfake."

It might be possible to layer a system with that amount of nuance on top of a blockchain, but you need to be really careful that your system won't end up making the world much worse before it has a chance of making it better.

Believable deep fakes, at least currently, cannot be created instantly. Blockchain guarantees whatever footage is on chain existed at the moment it was added and hasn't been tampered with since.

There are a lot of reputationally damaging images you could create for which it is not possible to determine the date when they were supposedly taken just from looking at them.

If you know that a specific cop is going to be at a specific location at a specific time, you could create the deepfake in advance and add it to the blockchain when they show up.

Sounds pretty abusable to me..

Maybe the cops should stop abusing citizens then.

More that a non-deletable video storage system will probably accumulate lots of spam.

So abusing innocent cops just doing their job is just collateral damage?

How would posting videos of good cops doing good things be "abuse"?

Because not every video has full context and not every use of force is unjustified?