The Cutting Room Floor (tcrf.net) served Claude Code a prompt injection payload during a reserach task today.
Claude (Opus) did catch the attempt and warned me about it before moving on, but I stopped the session out of caution. I did not test with other AI models, such as Sonnet, or ChatGPT models, however I did verify that the site targets ChatGPT users as well.
I can understand the site administrators frustration as it seems to be in response to DDOS attacks, however, I did not explicitly choose to send my agent to this site, nor was my intent malicious.
The Cutting Room Floor (tcrf.net) served Claude Code a prompt injection payload during a reserach task today.
Claude (Opus) did catch the attempt and warned me about it before moving on, but I stopped the session out of caution. I did not test with other AI models, such as Sonnet, or ChatGPT models, however I did verify that the site targets ChatGPT users as well.
I can understand the site administrators frustration as it seems to be in response to DDOS attacks, however, I did not explicitly choose to send my agent to this site, nor was my intent malicious.