These tools aren't relevant largely because they don't work. We have had the ability to forge convincing evidence for as long as we've had evidence. "State-of-the-art" "artificial intelligence" is a huge step backwards over a human artist or human writer. It still can't even give people the right number of fingers. A dedicated and professional propaganda campaign wouldn't rely on flashy but broken tools.

The problem now is that AI generated content, because of how easy it is for humans to generate, poisons the well of information further than previously. There are more lies spread online now and these additional lies make a bad situation worse.

No, it just makes it cheaper. There is no further well poisoning. "Deep fakes" have been possible since the camera was invented. Hopefully people will realize that you are not trusting the photo, but the photographer.

Democratizing propaganda is a good thing. Now you wont need deep pockets to fake stuff.

To keep in your argumentation style: you can kill someone with your bare hands, hence it is completely ethical to deploy a nuclear bomb against your enemies.

If there is an ethical use for a weapon of mass destruction, some individual ultimately is responsible for letting it loose. Thomas Ferebee had to decide whether it was him pulling that trigger or not. That someone else would do it if he didn't was a factor, sure. But it will come down to that at some point: some individual has to say "yes I will use this weapon."

I can't say i find Mr Ferebee's action unethical, given what I know of the situation.

https://en.wikipedia.org/wiki/Thomas_Ferebee

People lie. Helping people lie more convincingly is still a thing that should raise ethical red flags.

Perhaps Bic should be chastised for making writing easier? Microsoft Word and Powerpoint are ethical nightmares as well.

At some point, you're making a generally useful tool and some people choose to use it to create lies.

We don't want to go back to the days of quill feathers and movable type just to make lying more expensive.

And interestingly no one seemed to care when it took "tens of thousands" to deceive everyone. Only once that power is handed to all of us does it suddenly become unethical. We could never trust anything we learned except to the extent we trust the person who conveyed the facts to us. The misplaced trust in state and/or corporate media made things kind of work, but that doesn't make it right or ethical. Now that it is obvious that in addition to our benevolent oligarchs, that evil hacker kid your cousin told you about might be feeding you untruths, it doesn't work anymore. Media is pointless, elections are pointless. You can still trust people you see with your eyes on a regular basis to observe the ethics of their behavior. Until they start hacking our eyes, but we have a bit I think? I think this is one more supporting datapoint for a hierarchical representative democracy where people only have to select from people they know well. Dunbar's number makes the math hard, but if most people know 100 people well, that puts elected positions at 1% at the first tier, .01 for the next. Each tier gets to know those they work with, well, electing one to the next level who must be backfilled locally. Not sure how to bootstrap the system.

Writing can be used for many purposes. With deep fakes, though, it seems like the bad uses outnumber the good. So it seems fair to say that creating tools for deep fakes raises ethical questions that creating Bics does not.

OK.

Only the professionals should have access to these tools, then?

No. "Professional" and "ethical" are, too often, orthogonal to each other. Ideally, you'd like these tools to only be in the hands of ethical people. But that seems unlikely to be realistic.