Our finance team is losing their minds. Every time they run the month-end process in their old AS/400-based accounting suite, Intercept X's "deep learning" model flags it as a Trojan.
The software hasn't been updated in a decade. It's not malicious, it's just ancient and does weird things with memory. We've added exclusions, but the "behavioral" detection still trips. Support's answer is basically "add more exclusions or live with it."
Anyone else dealing with this? The marketing says "AI" prevents zero-day threats, but it seems more like it just freaks out over any legacy app that doesn't behave like a modern binary. Feels like we're tuning out the very protection we paid for.
How are you balancing this? Is the model trainable, or are we stuck whitelisting our entire legacy estate?
Trust but verify.