Within 6 months, ChatGPT could watch your screen, record calls: Sam Altman

OpenAI CEO Sam Altman says a future ChatGPT could watch screens, meetings, and calls, while connecting to emails, texts and documents to provide real-time context and assistance

OpenAI CEO Sam Altman has predicted a significant shift in how humans could interact with artificial intelligence (AI) assistants, saying a future version of ChatGPT could watch a user's computer screen, meetings, and record calls all the time.

 

In a recent interview with Cory Levy, founder of Z Fellows, Altman said, "I think we are close to a world where you can have a descendant of ChatGPT watch your computer screen all the time, watch every meeting you're in, record every call, everything that has perfect context of your whole life, everything you see."

 

When Levy asked when such a system could become useful, Altman gave a six-month timeline. "Some time within the next six months," he said.

 

 

Users would decide what information and services the AI could access. They could connect the assistant to text messages, email, documents, Slack or other sources, depending on the information they wanted to share. Altman said the system would not make decisions on behalf of users. Instead, it would work alongside them by maintaining context across their activities, offering suggestions and carrying out tasks.

 

He described a scenario in which a person, such as the CEO of a startup, has more work and information to manage than they can keep track of. The AI could work alongside the user as they type a sales pitch to a customer or write a strategy document. It could then suggest another idea, flag a potential mistake or offer to carry out a task.

 

"I think we're only one model generation away from this actually being incredibly useful," Altman said, adding that he hoped it would change the way he works.

 

AI that would understand users’ context

 

Altman described a proposed system that could use information shared by users to provide context-sensitive assistance.

 

For example, while writing a sales email, the AI could draw on customer history, past conversations and relevant documents to suggest improvements in real time. While drafting a strategy document, it could point out potential errors or offer alternative ideas based on the user's previous work.

 

The idea is that the assistant would not simply respond to individual prompts but understand what a user is doing across different activities and use that context to provide more relevant assistance.

 

ChatGPT’s existing memory features

 

OpenAI has already introduced features that allow ChatGPT to use information from previous conversations to provide more tailored responses. Users can manage memory and chat history settings, including turning off saved memories.

 

OpenAI's Computer History feature for the ChatGPT desktop app on macOS securely logs desktop interactions, including clicks, app switching, typing and keyboard shortcuts, to create a searchable timeline. The feature allows ChatGPT and Codex to recall workflow context without capturing raw audio or screenshots. It tracks event data such as app switches and clicks instead of taking continuous screen captures. The feature requires ChatGPT's Memories setting to be enabled so context can carry over across different tasks. It also automatically ignores activity inside private or incognito browsing windows and can suggest recurring workflows or automations based on repetitive habits.

 

The feature is strictly opt-in. Users choose which apps and websites the AI can track and can pause data collection or delete their stored activity history at any time.

 

Other tech giants such as Microsoft have developed a related approach through Recall, a Windows feature for Copilot+ PCs that captures snapshots of a user's active screen. Microsoft says Recall is opt-in, stores snapshots locally and encrypts them on the hard drive. Users can pause the feature, delete snapshots and control which content is captured.

 

The development of screen-level AI assistance has also raised security considerations. Earlier versions of Recall faced scrutiny after researchers identified ways to access stored snapshot data in preview builds. Microsoft subsequently added controls and security requirements around the feature.