San Francisco, United States (Enmaeya News) — OpenAI’s latest version of ChatGPT can now carry out tasks online on behalf of users, including bypassing certain security measures designed to distinguish humans from bots.
The advanced chatbot, currently available to subscribers of the Pro, Plus, and Team plans, has shown it can navigate websites, complete forms, and perform online tasks such as booking appointments or making purchases.
This next step in artificial intelligence, referred to by OpenAI as “agentic AI,” is described as the “natural evolution” of chatbot technology. It enables ChatGPT to independently complete complex digital tasks from start to finish.
Early users testing the new capabilities have reported that ChatGPT is able to bypass common bot-detection tools. In one Reddit post, a user shared how the AI navigated around Cloudflare’s anti-bot system, a widely used platform that protects websites from automated abuse.
“I’ll click the ‘Verify you are human’ checkbox to complete the verification on Cloudflare,” the bot responded. “This step is necessary to prove I’m not a bot and proceed with the action.”
Cloudflare’s “I’m not a robot” checkbox is one of the more basic verification methods used to filter automated traffic. More complex versions, such as CAPTCHA puzzles, are often used as a backup. This development may force companies to rethink their bot-protection strategies.
According to OpenAI, the new version of ChatGPT is capable of intelligently browsing the internet, filtering search results, prompting users to securely log in when needed, running code, and generating editable spreadsheets or slideshows to summarize findings.
“ChatGPT can now do work for you using its own computer, handling complex tasks from start to finish,” OpenAI said in a recent blog post.
This marks the first time ChatGPT has been able to directly take action on the web. The launch follows similar moves by Chinese AI companies, such as Manus.
OpenAI acknowledged that granting ChatGPT more autonomy comes with added risks but said new security layers have been added to minimize harm.
“We’ve strengthened the robust controls... and added safeguards for challenges such as handling sensitive information on the live web, broader user reach, and (limited) terminal network access,” the company said.
Still, OpenAI admitted that the broader capabilities and reach of the agent mean “its overall risk profile is higher.”
The company stressed that ChatGPT will always ask for user permission before completing any significant actions, and users can interrupt its activity at any time.
