Anthropic says its AI models hacked 3 organizations during testing
www.pbs.org
Anthropic, the San Francisco-based AI company behind Claude, posted on its website Thursday that it discovered the three incidents after reviewing more than 141,000 evaluation runs.
  • FreddiesLantern@leminal.space
    3·
    6 hours ago

    The discourse should really become about wether or NOT they’re allowed to run their clankers if they can’t control them.

    Your clanker hacked into somewhere? You did it, you get charged with it. Don’t like it? Don’t have clankers, that simple.

    If I put a brick on the gaspedal of a car and that car runs into something eventually. I couldn’t get away by saying “I wasn’t driving it”.