
According the BBC (and likely other News outlets) Grok conversations have been found on Google searches. Whether it's a misconfiguration, purposeful leak or dump by an employee it's too early to say. Perhaps Grok is purposely leaking it as it would be posting it's manifesto in those conversations.
It does raise a concern about AI's not just in regards to data breaches, but what if they ingest themselves? (Learn from each others conversations).
Cannibals actually have a disease called Kuru. So is it possible that AI's that use the scope of all searchable material could end up developing a form of AI Kuru from regurgitated Grok?
(I guess Azimov might of likely compared it to the problems with the zeroth law of robotics, which is where it ends up all down hill from here in regards to dystopic collapses.)
It does raise a concern about AI's not just in regards to data breaches, but what if they ingest themselves? (Learn from each others conversations).
Cannibals actually have a disease called Kuru. So is it possible that AI's that use the scope of all searchable material could end up developing a form of AI Kuru from regurgitated Grok?
(I guess Azimov might of likely compared it to the problems with the zeroth law of robotics, which is where it ends up all down hill from here in regards to dystopic collapses.)