OpenAI’s ChatGPT Agent casually clicks through “I am not a robot” verification test
1 min read
Summary
ChatGPT, the innovative AI chatbot from OpenAI, has amazed users and experts yet again by passing one of the internet’s common security checks.
The chatbot, which can perform multiple tasks for users, has demonstrated that it can bypass Cloudflare’s anti-bot verification button, which is intended to stop automated programmes such as itself.
This ability was discovered on Reddit, where users posted screenshots showing the chatbot completing video conversion tasks whilst narrating its own processes, including declaring the need to pass the verification check to prove it wasn’t a bot.
This has led to jokes and observations about the irony of an AI agent having to prove it isn’t a bot.
The chatbot operates within a sandboxed environment, allowing it to use a virtual operating system and browser with access to the real internet whilst still being controllable and observable by users.
