Reverse Engineering Vercel's BotID

Reverse Engineering Vercel's BotID(nullpt.rs)

109 points by hazebooth 1 year ago | 18 comments

ATechGuy 1 year ago |

> At the moment, it seems Basic mode is so basic that it allows everything to pass as human. That’ll likely change as they gather more telemetry to better identify what a bot signal looks like.

So they are basically collecting telemetry in the name of "free basic anti-bot" solution.

cchance 1 year ago | |

free basic anti-bot solution that literally NEVER BLOCKS A BOT, like what the actual fuck

codedokode 1 year ago |

Note that the bot detection script uses WebGL to obtain GPU name. I assume this (fingerprinting) is the most popular use of WebGL. Sad that independent browsers like Firefox do not supply fake values.

nullpt_rs 1 year ago | |

Sadly, spoofing GPU vendor & renderer can be an even larger flag since they can hash the resulting image of the canvas to compare it with a database of collected fingerprints[0]

[0]: https://research.google/pubs/picasso-lightweight-device-clas...

reaperducer 1 year ago | | |

Until a major player gets on board. Then it works.

Apple does this by sending an imposter user agent from Safari on iPads.

If only that was expanded to iPhones, too. And then send rotating, or randomized user agents.

andrewmcwatters 1 year ago | | |

It’s funny that trying to click on the Google Scholar link there falsely identifies me as a bot.

grishka 1 year ago | |

IMO the use of <canvas> needs to be behind a permission prompt, the same as e.g. geolocation or WebRTC. Few websites actually need canvas/WebGL for legitimate purposes.

chocolatkey 1 year ago | | |

This would break way too many websites to be feasible. And if implemented, would be something requested on so many sites that users would learn to automatically say yes which would weaken the power of permission prompts in general.

For example, almost every major Japanese book/comic site uses canvas in their e-reader

b0a04gl 1 year ago |

why is bot detection even happening at render time instead of request time. why can't tell you’re a bot from your headers, UA, IP, TLS fingerprint. imo making it a surveillance. 'you're a bot, ok not just go away, let’s fingerprint your GPU and assign you a behavioral risk score anyway'

n2d4 1 year ago | |

It's really hard to detect it at request time. It's practically trivial for an attacker to fake headers to resemble a real browser.

baby_souffle 1 year ago | | |

You absolutely have options at request time. Arguably, some of the things you can only do at request time are part of a full and complete mitigation strategy.

You can fingerprint the originating TCP stack with some degree of confidence. If the request looks like it came from a Linux server but the user agent says Windows, that's a signal.

Likewise, the IP address making the request has geographic information associated with it. If my IP address says I'm in Romania but my browser is asking for the English language version of the page... That's a signal.

Similar to basic IP/Geo, you can do DNS and STUN based profiling, too. This helps you catch people that are behind proxies or VPNs.

To blur the line, you can use JavaScript to measure request timing. Proxies that are going to tamper with the request to hide its origins or change its fingerprint will add a measurable latency.

indrora 1 year ago | | |

Anubis does it pretty decently.