When an LLM outputs, “I panicked”, it does not mean it panicked. It means that based on the preceding sentences, “I panicked” was a likely thing to come next.
It means it’s read a lot of fiction, in which drama is necessary.
It didn’t “panic”. It didn’t *anything*. It wrote a likely sequence of words based on a human request, which it then converted into code that matched those words somewhat. And a human, for some reason, allowed that code to be evaluated without oversight.
ATTENTION TO TAILSCALE USERS
Tailscale has recently updated their terms of service to include a forced arbitration clause and class action waiver. YOU HAVE 30 DAYS TO OPT OUT OF THIS. If you do not opt out, Tailscale will automatically assume that you've accepted the terms, and you will be unable to sue Tailscale or participate in a class action against them in the future. During arbitration, YOU will be required to pay the cost of initiating arbitration.
Opt out notices must be sent to Tailscale by mail. However, Tailscale is vague about which address to send to, since they have an address in Canada and in California listed.
Once again, lawsuits are the only recourse we have against companies when they fuck up, we cannot let them take away our right to sue, or else there is nothing to hold them accountable.