The settlement should include the removal of anything that was gained by training on the data. Meta will complain that the data is too ingrained in the model and can’t be removed. They likely aren’t wrong but that does not seem like Blacked’s problem. Maybe sell them a license to continue to use the data at $200k per year per video until they can definitively prove that none of the data is still in the model?
Now that would be interesting, because you can’t prove a negative. The only way for Meta to prove that none of the videos is making their way intot he model is to provide clear evidence of all the other videos and stuff that are making their way into the model.
Now that would be interesting, because you can’t prove a negative. The only way for Meta to prove that none of the videos is making their way intot he model is to provide clear evidence of all the other videos and stuff that are making their way into the model.
The settlement should include the removal of anything that was gained by training on the data. Meta will complain that the data is too ingrained in the model and can’t be removed. They likely aren’t wrong but that does not seem like Blacked’s problem. Maybe sell them a license to continue to use the data at $200k per year per video until they can definitively prove that none of the data is still in the model?
Now that would be interesting, because you can’t prove a negative. The only way for Meta to prove that none of the videos is making their way intot he model is to provide clear evidence of all the other videos and stuff that are making their way into the model.
Now that would be interesting, because you can’t prove a negative. The only way for Meta to prove that none of the videos is making their way intot he model is to provide clear evidence of all the other videos and stuff that are making their way into the model.