Johann Wilfred
- OpenAI just debuted a new ChatGPT agent capable of carrying out complex tasks using multiple websites
- The feature is now available for ChatGPT Pro, ChatGPT Plus, and ChatGPT team users
- Sam Altman warned on X that the feature could be misused by malicious actors
What started as an initial tease on X of a video alerting ChatGPT users about a potential new product finally rolled out in the form of a new feature called ChatGPT Agent.
An extended addition for ChatGPT Plus, ChatGPT Pro, and ChatGPT team users, OpenAI’s new agent is able to browse the web and conduct complex, detailed tasks typically done over a timeline. The agent will also be available to Enterprise and Edu users in the coming weeks.
On Thursday, OpenAI’s X page put out a promo video of a cursor hovering over a blank page before finally landing on a message that ran “I’m tuning into the livestream tomorrow at 10am PT”
The official unveiling of the feature took place on Friday when Sam Altman and a team of AI experts—Casey Chu, Isa Fulford, Yash Kumar, and Zhiqing Sun—sat down for a video demo of how the agent would work in certain situations. ChatGPT agent demonstrated an ability to be able to solve complex tasks such as planning a visit for a wedding event, along with outfit suggestions, hotel bookings, and an ideal wedding gift.
However, the team warned about the possibility of malicious websites using the AI agent to harm the user, a statement echoed by Sam Altman on X as well when he announced the new agent.
“We don’t know exactly what the impacts are going to be, but bad actors may try to “trick” users’ AI agents into giving private information they shouldn’t and take actions they shouldn’t, in ways we can’t predict. We recommend giving agents the minimum access required to complete a task to reduce privacy and security risks.” said Altman in the X post
