That person’s take on “AI CSAM” also misses the point.

There are multiple issues.

1) A company doesn’t communicate well how a product actually works. In fact, I’m not sure they have much insight into that. In the end, someone ends up Blaming The User based on how it might work. There are quite a few guesses made based on the worst scenario someone can imagine.

2) A deepfake is not bad because it’s offensive, but that is often how it is portrayed. It’s bad because of it being a misuse of someone’s likeness impinging on their privacy. Too much chatter talks about something through the lens of it being content as if it’s about a bit of content being offensive.

A focus on “content” leads to bad policies which don’t help and actually makes it harder to deal with the issue.

3) It misses how it plays into broader copyright / data protection concerns.

4) AI is a technology, so someone can use it more or less ethically. The label “AI CSAM” assumes any use of AI is unethical. This leads to bad policy.