News for hackers. Stuff that matters.
Re:Google Open-Sources Trillion-Parameter AI Language Model ... by airocker on Wednesday February 17, 2021 @08:19PM (#26174251)
Would GDPR or other regulation apply to the data that these model is trained with? Is it not a risk that the model will record some private information in someone's email?
Re:Google Open-Sources Trillion-Parameter AI Language Model ... by ma2rten on Wednesday February 17, 2021 @09:50PM (#26174886)
Yes, it has been shown [1] that these models can memorize information even if it only appears once in the training data. It could potentially cause issues with privacy and especially copyright. However, in this case it's not relevant because they didn't release the model weights.

[1] https://arxiv.org/abs/2012.07805

Re:Google Open-Sources Trillion-Parameter AI Language Model ... by airocker on Wednesday February 17, 2021 @10:47PM (#26175457)
Even if they did not, they have it for themselves. If this uses any personal data, then GDPR and other regulations may not be getting respected.
Re:Google Open-Sources Trillion-Parameter AI Language Model ... by visarga on Thursday February 18, 2021 @01:06AM (#26176517)
If they have a policy to respond to GDPR requests, and maybe if they use a bloom filter to avoid repeating the training data, then it should be ok?
Re:Google Open-Sources Trillion-Parameter AI Language Model ... by bidirectional on Thursday February 18, 2021 @08:44AM (#26179694)
I feel like a future version of GPT will easily be able to answer 'What is the home address of [email protected]', given the abundance of data it is trained on and the widespread proliferation of leaked databases.
Re:Google Open-Sources Trillion-Parameter AI Language Model ... by neximo64 on Wednesday February 17, 2021 @09:33PM (#26174767)
Maybe host and use it in a non GDPR environment, it's always going to be ambiguous and risky.