The moral of the story of the Meta AI person getting their email deleted by clawdbot is not “haha, so funny” or even “haha, so irresponsible in their position” but “it’s normal to be reckless with AI and it screwing up” in order to get you more comfortable giving these programs access to your most personal data. Do you think it would have gotten past Meta PR any other way? Don’t fall for it.
Related: ever since 2010, when Path popularised the “upload your address book to bring in your social graph” and Apple in response _finally_ requiring a user prompt for access to the address book I’ve been mulling over the problem of me giving you my contact details and then you handing them over without my consent.
The only solution I can come with is DRMing each record and have the OS enforce that consent.
Never gonna happen, but I’d love for someone to build consent into data sharing.
@janl ideally same for photos taken: 'automate the "I'm happy to be in your archive, but not on Facebook let alone shared in a public timeline" decision'
