We have automated account creation and probing against an application on AWS. The traffic has adapted to rate limits, tougher CAPTCHAs, and IP blocks, including blocks on Tor exit nodes.
What defenses are worth trying when the traffic keeps changing addresses and tactics?
_p61+vexlato.9:
Keep the raw request details needed to investigate, but avoid logging passwords or other sensitive values. A pattern is more useful if you can tell which accounts and actions it connects.