The OpenAI Operator leak suggests it’s coming to the ChatGPT Mac app soon – here’s why it’s so important
Aside from the possible introduction of artificial general intelligence (AGI), AI agents, autonomous processes that you can instruct to perform complex tasks for you on your computer, may be the biggest new AI feature in 2025. Agents can be essential for changing your computer. mobile phone into a real AI assistant, which can do what you ask it to do without you having to get involved.
OpenAI has been teasing us for a while with the release of its first AI agent, called Operator, but the latest code leak suggests it could be coming to the Mac very soon.
A new leak on X by Tibor Blaho claims to have revealed evidence that OpenAI’s Operator agent is coming to the ChatGPT Mac app. Tobor has discovered hidden options to define desktop launcher shortcuts to ‘Toggle Operator’ and ‘Force Quit Operator’, which suggests you might need a quick way to shut it down if things get out of hand!
Confirmed – the ChatGPT macOS desktop app has hidden options to define desktop launcher shortcuts to “Toggle Operator” and “Force Quit Operator” https://t.co/rSFobi4iPN pic.twitter.com/j19YSlexASJanuary 19, 2025
Tobor also claims to have found code in the browser version of ChatGPT that refers to Operator with references to an ‘Operator System Card Table’, ‘Operator Research Eval Table’ and ‘Operator Refusal Rate Table’. The last entry indicates that the Operator may fail to perform the tasks asked of it, enough to require a decline rate.
The OpenAI website already contains references to Operator/OpenAI CUA (Computer Use Agent) – “Operator System Card Table”, “Operator Research Eval Table” and “Operator Refusal Rate Table” Includes comparison with Claude 3.5 Sonnet Computing, Google Mariner, etc . (example of tables… pic.twitter.com/OOBgC3ddkUJanuary 20, 2025
Automation of daily tasks
Recently, one of OpenAI’s founders, Wojciech Zaremba, criticized rival Anthropic in a message on X for releasing its AI agent without the necessary security measures in place. His message read:
“Anthropic – just released a computer-using agent without any security restrictions. I can only imagine the negative reactions if OpenAI made a similar release.”
Where does the gap between perception and reality in AGI company safety come from? https://t.co/hu2UBxwkSW — Elon has been very vocal about security, but so far no one at https://t.co/hu2UBxwkSW is working on safety.Anthropic — just released a computer-using agent without any security…December 14, 2024
It’s the ability of an AI agent to integrate into your daily computing tasks that makes it such a big step forward for AI and has the potential to completely change the way we interact with our devices. Imagine if you didn’t have to book hotel rooms, pay bills, or even write code.
Clearly, a lot of work needs to be done before people will trust an AI agent to perform such tasks autonomously, and privacy will be a key issue.