Sam Altman Says ChatGPT Could Soon Watch Your Screen and Meetings

OpenAI CEO Sam Altman has outlined a future in which ChatGPT becomes far more proactive, potentially observing what users are doing on their computers and understanding the context of their work. Instead of waiting for someone to type a question, an AI assistant could continuously understand what is happening and offer useful assistance when needed. Recent discussion around Altman’s comments suggests this type of capability could arrive within roughly a single model generation.

One example Altman has highlighted is meetings. An AI assistant could potentially follow a meeting, understand the discussion and use that context to help with related tasks. Rather than functioning only as a chatbot, the system could become a digital assistant capable of understanding conversations, documents and other information surrounding a user’s work.

The idea represents a major shift in how people could interact with AI. Today’s assistants generally require users to provide explicit prompts, but a context-aware system could recognize what a person is working on and provide assistance without being asked every time. That could include helping organize information, summarizing discussions or identifying relevant details while someone works.

However, technology capable of continuously observing screens or meetings would also raise significant questions about privacy and user control. People would need clear choices about what an AI can access, when it can observe information and how that information is stored or processed. The distinction between an assistant that helps when requested and one that continuously monitors activity would be particularly important.

Altman’s vision reflects the broader AI industry’s movement toward agentic AI, where systems can understand context and take useful actions rather than simply generate responses. If the technology develops as predicted, ChatGPT and similar assistants could become deeply integrated into everyday digital workflows. The challenge for OpenAI will be making these systems genuinely useful while ensuring users remain in control of what their AI can see and do.