OpenAI's ChatGPT agent may pose data theft risks, warns Altman
OpenAI CEO Sam Altman is sounding the alarm about their new ChatGPT Agent.
He called it "cutting-edge and experimental" but warned that criminals might try to use it to steal sensitive info.
His advice? Don't use it for anything high-stakes or share your personal data with it, just to be safe.
About the ChatGPT agent
This tool lets you do some pretty advanced stuff online using AI and a remote visual browser.
To keep things safer, OpenAI has added features like clear user consent, special supervision for risky tasks, and an option to delete your session data.
OpenAI is working with governments and security experts
Altman says tackling these risks takes teamwork between tech and society.
That's why OpenAI is working with governments and security experts through red team testing and a bug bounty program—basically inviting people to help spot problems—so they can keep improving safety for everyone using the ChatGPT Agent.