"It also produced more false positives than humans, sometimes mistaking harmless network signals for signs of a successful breach. The researchers noted that th

"It also produced more false positives than humans, sometimes mistaking harmless network signals for signs of a successful breach. The researchers noted that the system performs best in environments dominated by text-based inputs and outputs." How many more? The article doesn't say.