Why Building an AI Kill Switch Is So Hard

Bullish Gator

Experts explain technical, hardware, and regulatory hurdles—plus EU AI Act and self‑regulation limits—that make designing a reliable AI kill switch difficult and possibly impractical.

Editorial companion

What stands out

CNBC gathers experts to show why a single, reliable AI "kill switch" is far harder to build than headlines suggest. The clip lays out practical hurdles: models operate across software, firmware, and physical chips, hardware shutoffs can be bypassed or misused, and enforcement across suppliers and jurisdictions is technically and politically complex. It also flags regulatory limits, the EU AI Act and industry self-regulation leave gaps, so a tidy emergency cutoff looks unlikely without major technical redesigns and international cooperation.