Leaks hint at Operator-like tool in ChatGPT ahead of GPT-5 launch
A few new code references in the ChatGPT web app and Android point to an Operator-like tool in GPT’s chain of thoughts.
As spotted by Tibor on X, the Android beta has strings like “click,” “drag,” “type,” and even “terminal feed,” which seem to suggest that ChatGPT could soon call a remote browser or sandboxed environment (Operator?).
For those unaware, OpenAI already has Operator, which uses an AI agent to navigate a remote browser session and execute tasks for you.
These references suggest that ChatGPT’s thinking model might call a browser or APIs to execute tasks. There are also references to “Checking available APIs” and “Reading API documentation.”
It’s also worth noting that the code mentions an “intake form,” so OpenAI may gate this feature behind an invite-only beta before rolling it out more widely.
Could it be ChatGPT-5 or some other model? We don’t know, and we likely won’t anytime soon as OpenAI is busy with its standoff with Meta.
While cloud attacks may be growing more sophisticated, attackers still succeed with surprisingly simple techniques.
Drawing from Wiz’s detections across thousands of organizations, this report reveals 8 key techniques used by cloud-fluent threat actors.
Source link
