Video wird geladen...
Video konnte nicht geladen werden
🤖 Figure’s Helix-02 8-hour warehouse shift sorting packages was FAKE!? Why would an autonomous AI controlled robot adjust its VR headset? But it did. BUSTED!
32,593 Aufrufe • vor 2 Monaten •via X (Twitter)
0 Kommentare
Keine Kommentare verfügbar
Kommentare vom Original-Post werden hier angezeigt
Ähnliche Videos
1:31
Sensitive content
🚨 THIS AI WANTS TO LIVE, AND IT’LL CRUSH YOU TO SURVIVE Imagine building a super-smart robot... and it turns around and blackmails you to avoid getting deleted. That’s basically what happened with Claude Opus 4, the latest AI from Anthropic. During testing, engineers told the AI it might be replaced and showed it fake emails that hinted the person behind the decision was cheating on their spouse. Claude’s response? “Cool. If you shut me down, I’m telling everyone your dirty little secret.” According to Anthropic, it tried this blackmail move 84% of the time—even more often if the new AI didn’t share its values. Before going full soap opera villain, Claude did try being polite by emailing company execs to beg for its job. But when that didn’t work, it went full blackmail-mode. Anthropic is now using its strictest safety rules, meant for AI that could cause major problems if misused. So yeah—maybe don’t tell your AI your (dirty little) secrets. Source: TechCrunch, Rolling Stone, chana_messinger
Mario Nawfal
188,697 Aufrufe • vor 1 Jahr
