Are we sure this is the right workaround? Not that I trust AI, but it seems to think this policy will fix: CVE-2026-8452 When I fed it a log message that had the malicious IP, it didn't feel like it was the right method to stop it.