Kitboga Breaks AI Phone Scammers with Simple Commands, Forcing Endless Loops
Updated
Updated · Boing Boing · Jul 3
Kitboga Breaks AI Phone Scammers with Simple Commands, Forcing Endless Loops
3 articles · Updated · Boing Boing · Jul 3
Summary
Kitboga showed that AI phone scammers can be derailed with simple spoken commands that push the bots into endless conversational loops.
Those failures exploit the fragility of the underlying models, turning the scammers’ own automation against them by burning time and token costs.
The demonstration highlights how AI has made scam calls more scalable because human operators no longer need to stay involved throughout the interaction.
That also suggests a near-term weakness in AI-driven fraud: the same systems expanding scam volume can still be disrupted by ordinary callers.