Clearly, given there have been cases about it already.
> Also, it seems you are fundamentally missing how AI is different. What would you expect a “regulator mandatory correction” to look like, a one sentence summary comes with a notice that it was corrected at some point?
Laws shouldn't care about mechanisms, they should care about outcomes. We can't allow ourselves to say "this tech makes ${legal obligation XYZ} impossible to perform, let us deploy it anyway": either you figure out how to solve for ${XYZ} or you don't get to deploy it.
> AI is also going to make regulators and bureaucrats totally superfluous if done properly, where AI simply “regulates” based on laws written in a clear text and open weight manner.
"If" is doing heavy lifting there — "if done properly", AI makes all human labour redundant. Nobody knows how far away "done properly" is.
Is libel in AI generated summaries a problem?
Also, it seems you are fundamentally missing how AI is different. What would you expect a “regulator mandatory correction” to look like, a one sentence summary comes with a notice that it was corrected at some point?
AI is also going to make regulators and bureaucrats totally superfluous if done properly, where AI simply “regulates” based on laws written in a clear text and open weight manner.