Watch Desk posted an update
US lawmakers are pushing for emergency switches to shut down AI, but experts warn that disconnecting an advanced model may not be straightforward, Bloomberg reports.
Why it mattersThe tension is the point: a switch is only a safeguard if it can reliably stop the system when needed. The debate puts a practical question behind the familiar call for human oversight: what would an emergency stop actually do?
Discuss: Should advanced AI systems be required to have a reliably testable emergency shut-off, even if some systems cannot be stopped with a single switch?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.