Bots generating errors on my website

[Corobori]

I suddenly noticed hundreds of errors on the log file for a website I created that are all caused by openai bots as the HTTP_FROM says gptbot(at)openai.com and USER_AGENT is Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.2; +https://openai.com/gptbot)

I looked into one example and it seems the path created goes way out of bound. This is the path /Odonto-lieyectoresli-541.aspx/assets/js/plugins/Docs/Productos/assets/js/Docs/Productos/assets/js/assets/js/assets/js/vendor/images2021/Docs/Productos/Docs/Productos/assets/js/vendor/Docs/Productos/Docs/Productos/Docs/Menu/Odonto-Gomas-para-pulido-de-composite-tipo-Enhace-1815.aspx

Before blocking the bot I am writing to find out what’s wrong with my website.

[neofutur]

Same problem here, with very long URLS with infinite number of &amp:
20.171.206.213 - - [29/Oct/2024:10:56:16 +0100] “GET /298-bois-chene-moderne?amp%3Bamp%3Border=product.price.desc&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Classique+Chic&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Classique+Chic&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Classique+Chic&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bamp%3Bq=Style-Design&amp%3Bamp%3Bq=Style-Design&amp%3Border=product.name.desc&order=product.name.desc HTTP/1.1” 200 25215 “-” “Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.2; +https://openai.com/gptbot)”

Please FIX THIS or stop crawling.

[Corobori]

Glad to see I am not the only one.

More than 4000 errors today, I am considering blocking the bot thru robots.txt

[neofutur]

Same here, I took the time to come here and report the problem, but I fear chatgpt developpers wont read us, there is no “BUG REPORT your AI is broken , please fix this” ticketing system . . . If we have no answer qite fast, I ll also block OpenAI . robots.txt or better . . . firewall dropping the whole IPrange NetRange: 20.160.0.0 - 20.175.255.255
CIDR: 20.160.0.0/12
NetName: MSFT

In the end I had no choice , those gptbot attacks were overloading my dedicated server, so I had to reject all gptbot requests :

ufw insert 1 deny from 20.171.206.0/24

for now. until openAI fix the buggy pseudo AI.

[Corobori]

From my side all errors reported above are not showing up anymore. I don’t know if something was fixed or if the crawler isn’t coming around anymore

[PaulBellow]

Thanks for coming back to let us know!

[dlh.dam]

Hi same thing form me i’m in france,
the bot come every 3 days and my website show error 500…

[krefects]

Estou com o mesmo problema.
mesmo bloqueando pelo robots.txt, ainda continua a derrubar meu site.
agora vou aguardar 24 horas para ver.

/midias/?amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;page=6&page=4&page=6&page=6&page=5&page=6&page=5&page=3&page=3&page=5&page=4&orgao=8&page=5

[krefects]

# Bloquear IPs e faixas de IP
RewriteCond %{REMOTE_ADDR} ^52\.230\.152\. [OR]
RewriteCond %{REMOTE_ADDR} ^52\.233\.106\. [OR]
RewriteCond %{REMOTE_ADDR} ^20\.171\.206\. [OR]
RewriteCond %{REMOTE_ADDR} ^20\.171\.207\. [OR]
RewriteCond %{REMOTE_ADDR} ^4\.227\.36\.([0-9]|[0-9][0-9]|1[0-1][0-9]|12[0-5])$ [OR]
RewriteCond %{REMOTE_ADDR} ^20\.42\.10\.1(7[6-9]|8[0-3])$ [OR]
RewriteCond %{REMOTE_ADDR} ^172\.203\.190\.1(2[8-9]|3[0-9])$ [OR]
RewriteCond %{REMOTE_ADDR} ^51\.8\.102\. [OR]
RewriteCond %{REMOTE_ADDR} ^189\.89\.12\.150$
RewriteRule ^ - [F,L]

[neofutur]

looks like someone understood how to use chatgpt to ddos websites, and no one cares at (Open)ClosedAi.

[drmike]

I have the same problem - GPTBot/1.2 is hitting nonexistent endpoints on my site thousands of times each day, filling up my logs and depleting my Rollbar credits.

[jochenschultz]

Are you sure it is GPTBot? Did you check the IP?

[drmike]

Good question.

Yeah, the IP is OpenAI.

It was 20.171.207.223, which is listed in https://openai.com/gptbot.json (as “20.171.207.0/24”)

whois says that block is owned by Microsoft, so I suspect OpenAI is running the bot on Azure:
NetRange: 20.160.0.0 - 20.175.255.255
CIDR: 20.160.0.0/12
NetName: MSFT
NetHandle: NET-20-160-0-0-1
Parent: NET20 (NET-20-0-0-0-0)
NetType: Direct Allocation
OriginAS:
Organization: Microsoft Corporation (MSFT)
RegDate: 2017-02-22
Updated: 2017-02-22

[jochenschultz]

And the pages never existed?

[drmike]

Let me be more precise - the “GET” endpoint never existed, only “POST” endpoints for forms.

Users (and GPTBot) are served form code that begins like this:

<form data-turbo="true" method="post" action="/twitter/twitter_user/261377/upvote">

Apparently GPTBot scrapes that URL and tries to HTTP “GET” it, which is meaningless in the context of the application. My Rails server was serving a 404 for it.

I’m not clear on why the bot decided to slam those forms with thousands of requests

[jochenschultz]

@vb could you look into this?

[colinr]

Thanks for flagging this, we’re looking into it. We do monitor gptbot(at)openai.com if you’d like to reach out directly.