Botid false positive, what now?

Help
1

Hi @quuu :waving_hand:

We are now running with BotID in production, with the deep analysis enabled for our login route.

Works great for most users, but… Today I talked with a customer, she could not get in.

I traced our logs, here it says:

"verification": {
      "bypassed": false,
      "isBot": true,
      "isHuman": false,
      "isVerifiedBot": false
    }

I also tracked this event in Sentry, where I have some more details:
User
Geography Dublin, Ireland (IE)

Browser
Chrome 139

Runtime
node v22.15.1

Operating System
Kernel Version 5.10.239-261.959.amzn2.x86_64
Name Linux

Client Operating System
Name Windows

What now? Why is she being detected as a bot? What are best practices in a case like this?

BotID Incorrectly Flagging All Traffic as Bot
3

hi @alexbjorlig , just confirming - is this for 21risk?

the most immediate thing if a user has reached out about this is to create a firewall bypass rule to let your user through WAF System Bypass Rules

I’ve looked through the logs of botid from your project and I can see some sessions being flagged by our partner Kasada as bots. Can you give me a rough time stamp of when the classification happened? will help narrow down the exact session and reason

1 Like
4

Thanks for looking into this :+1:

Yes, it’s 21RISK. RequestId was lhr1::c4cck-1756987608330-fdfd3db13a58 at Sep 4, 2025 14:06:48 CEST.

Regarding the WAF rule, something like this?

CleanShot 2025-09-04 at 14.51.42@2x
CleanShot 2025-09-04 at 14.51.42@2x1516×1710 109 KB

5

Not hostname because it’d bypass more than just that user, if that user only accesses your site from a certain ip, or ASN or user agent, you should bypass that

Otherwise in code you can add a feature flag like

If (user.email===…)

Break;

// only if user is not that email, then

CheckBotID()

Sorry for bad formatting, was on mobile :stuck_out_tongue:

6

Ok, we already deployed that as a hotfix.
But what is the underlying challenge - why was the user identified as a false positive?

7

Dug deeper and figured it out - that user is using ZScalar, and it’s acting like a reverse proxy and obscuring signals we rely on for bot protection.

I’m working with the security team to detect usage of VPN’s and not block ZScalar users - but this is a problem with any reverse proxy or zero trust proxy in front.

1 Like
8

Thank you so much for sharing this insight with us, it’s great to know why. Not that we can do much about it, but you know :smiley:

Hopefully the strategies can be improved, so legit users are not detected as bots in the future

2 Likes
9

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.