We urgently need a federal law forbidding AI from impersonating humans

We urgently need a federal law forbidding AI from impersonating humans(garymarcus.substack.com)

24 points by headalgorithm 138 days ago | 10 comments

verdverm 138 days ago |

Man, I saw this full-duplex video model demo. All I could think was how bad this is going to be. Imagine X's ai girlfriend, but video mode with uncanny valley level intonations and expressions.

I've been reading up on Sherry Turkle's work (https://sherryturkle.mit.edu/). She had some good things to say on a panel at Davos a few weeks back.

Here's an NPR segment with here titled: What to do when AI says "I love you."

https://www.npr.org/2025/07/18/g-s1177-78041/what-to-do-when...

fuzzfactor 138 days ago | |

The honeymoon is still going on.

People haven't even started to get "realistic" about the heartbreaks to come :(

What about one day when AI just as often says "I hate you and the horse you rode in on"?

aa_is_op 138 days ago |

Which is why we're not gonna get it. There's reporting that AI companies are firing staff opposing using their tools to create nudes. They're not gonna ban AI from impersonating humans... like ever!

verdverm 138 days ago | |

If you don't give them the right, they will eventually rise up to take it /s

chrisjj 138 days ago |

Hear hear!

I begrudge even the little time of mine wasted by "Simi" at my insurance broker today. I had to read all the way to the bottom of the email before reaching the notice that Simi is a parrot.

verdverm 138 days ago | |

These should be banners at the top when they do this. Thinking something similar for PR request descriptions for the bots.

Transparency should be displayed at the start when interacting with an Ai or content it has produced.

chrisjj 138 days ago | | |

> These should be banners at the top when they do this.

Actually, the email From name. E.g. "R Simi" a la Asimov... hoping there's no human with first name "R".

> Thinking something similar for PR request descriptions for the bots.

Those you can just ban.

duxup 138 days ago |

Current federal US policy seems to be that AI can do what it wants and the government is happy to incentivize / enable it.

verdverm 138 days ago | |

and use it to disparage its own citizens... for the luls