writearthur.bsky.social
Friendly reminders about spooky technologies.
Words in Wired, The Economist, The Atlantic, and others. Author of “Eyes in the Sky”
166 posts
226 followers
676 following
Prolific Poster
Conversation Starter
comment in response to
post
If the law had a carve-out for videos of "public figures," then a platform could leave the video up. But that means that it could also leave up a deepfake of Taylor Swift, or AOC.
It's a messy and unsettling problem. And I don't think people are talking about it enough.
comment in response to
post
Yes, social media platforms are not bound by the First Amendment, but any law that compels platforms to remove depfakes would apply to the Trump/Musk video. Meaning that many of us would have never ended up seeing the video.
comment in response to
post
Just like thirsty guys, thirsty tech is annoying at best, very dangerous at worst, and totally sad across the board.
comment in response to
post
yes, it's a similar thing. but emails and blogs are also different in many ways.
comment in response to
post
But also, will people be less polite? More polite? Will they use fewer exclamation points? Will they still say "I hope this finds you well" or "as per my previous email" ? How will they try and game the AI?
comment in response to
post
Of course, if the OPM emails end up leaking onto the open Web, that's a different story.
comment in response to
post
It's a really good question. Though as I understand it, input data does not automatically train models. So unless xAI deliberately puts the OPM emails in a training dataset for the next Grok update, ChatGPT would not be training Grok in this instance.
comment in response to
post
*Not asking because this has happened to me. (At least as far as I know). I’m just curious.
comment in response to
post
We also assembled a list of detailed questions that you can ask your city government or police department if they are currently using this technology, or if they are planning to use it in the future. accelerator.carnegiecouncil.org/data-fusion/...
comment in response to
post
Last year @hypervisible.bsky.social and I put together a tool showing how fusion tech like Fusus enables precisely this type of behavior, among many other risks. accelerator.carnegiecouncil.org/data-fusion/