TikTok is being flooded with racist AI videos generated by Google's Veo 3

1. ivape ◴[02 Jul 25 23:46 UTC] No.44450061[source]▶

I think it's fine to fingerprint AI generated images/videos. It's a massive privacy violation but I just can't see any other way. Too many people have always been and will always be unethical.

replies(3): >>44450255 #>>44450415 #>>44450639 #

2. SchemaLoad ◴[03 Jul 25 00:17 UTC] No.44450255[source]▶

>>44450061 (TP) #

I've been wondering if ChatGPT makes such excessive use of EM dash just so people can easily identify AI generated content.

Google wouldn't even need a fingerprint, they could just look up from their logs who generated the video.

replies(2): >>44450363 #>>44450962 #

3. oceanplexian ◴[03 Jul 25 00:35 UTC] No.44450363[source]▶

>>44450255 #

Google already admitted they are fingerprinting generative video and have a safety obsession so I guarantee they do it to their LLMs. Another reason is to pollute the output that folks like Deepseek are using to train derivative models.

4. partiallypro ◴[03 Jul 25 00:43 UTC] No.44450415[source]▶

>>44450061 (TP) #

Eventually as models become cheaper, the big companies that would do this won't have control over newer generated content, so it's fairly pointless.

5. WillPostForFood ◴[03 Jul 25 01:18 UTC] No.44450639[source]▶

>>44450061 (TP) #

To what end? You want to fingerprint all AI images and video to catch people who make racist videos in order too to do what? It isn't illegal. If TikTok doesn't like the content they can delete the video and the account. If Google or OpenAI doesn't want the content being created, they can figure out a way to block it, and delete the user's accounts in the meantime.

If I told you many 14 year olds were making very similar offensive jokes at lunch in high school, would you support adding microphones throughout schools to track and catch them?

replies(1): >>44451568 #

6. IAmGraydon ◴[03 Jul 25 02:11 UTC] No.44450962[source]▶

>>44450255 #

The em-dash is one marker, but I’ve read that most LLMs create small but statistically detectable biases in their output to help them avoid reingesting their own content.

7. ivape ◴[03 Jul 25 04:13 UTC] No.44451568[source]▶

>>44450639 #

If I told you many 14 year olds were making very similar offensive jokes at lunch in high school

A picture is worth a thousand words. Me saying your mom is so fat that _______ in the lunchroom is different than me saying your mom is so fat in cinematic video format that can go locally viral (your whole school). This is the first time in my life I'm going to say this is not a history is echoing situation. This is a we have entirely gone to the next level, forget what you think you know.