OpenAI CEO Sam Altman has said a future generation of ChatGPT could gain the ability to watch a user’s screen, record meetings and calls, and build a much broader understanding of the user’s activities within roughly six months. The comments point toward a more proactive form of AI assistant that would observe what users are working on and provide assistance using that real-time context, rather than waiting for users to manually upload information or ask questions.
Altman’s comments were made at an OpenAI intern event and describe a potential future capability rather than a confirmed product launch or release schedule. The proposed system could connect information from screens, meetings, calls, emails, texts and documents to create what Altman described as highly comprehensive personal context. Such capabilities could make ChatGPT substantially more useful for work and productivity, but they also raise significant questions about consent, privacy, data security and the boundaries of always-on AI assistance.
ChatGPT Could Watch Screens Within Six Months
Altman suggested that OpenAI could be only about one model generation away from a ChatGPT system capable of observing what a user is doing on their screen and using that information to provide assistance.
The concept is different from today’s typical chatbot interaction. Instead of explaining a problem by typing a prompt or uploading a document, a user could potentially allow ChatGPT to see the work already happening on their computer.
That could include writing a sales presentation, working on a strategy document, reviewing data or navigating software.
Potential Future Capabilities
| Capability | Potential Use |
|---|---|
| Screen access | Understand what the user is currently working on |
| Meeting recording | Summarise discussions and identify action items |
| Call recording | Capture important points from conversations |
| Email access | Understand ongoing communications |
| Text access | Add personal conversational context |
| Document access | Work with existing files and information |
| Persistent context | Provide more personalised assistance |
These are capabilities Altman described as part of a possible future ChatGPT experience. They should not be interpreted as a confirmed list of features that OpenAI has committed to shipping within six months.
From Chatbot To Context-Aware AI Assistant
The significance of Altman’s comments is not simply that ChatGPT could see a screen. The larger shift would be from an AI that responds to individual prompts toward an assistant that understands the broader context surrounding a user’s work.
Today, a user generally has to provide context manually.
A future system could potentially gather some of that context automatically, assuming the user grants the required permissions.
Evolution Of AI Assistance
TODAY
User
↓
Types Prompt
↓
Provides Context
↓
ChatGPT
↓
Generates Answer
POTENTIAL FUTURE
User
↓
Screen + Meetings + Calls
+ Emails + Documents
↓
Continuous Context
↓
ChatGPT
↓
Proactive Assistance
This would allow AI to respond not only to what a user says but also to what the user is actively doing.
What Could Screen-Watching ChatGPT Do?
The practical applications could be extensive.
A user preparing a presentation, for example, could potentially receive suggestions while working rather than after completing the document and asking ChatGPT for feedback.
A developer could potentially ask ChatGPT to identify an error visible on screen without manually copying the error message. A financial analyst could receive help interpreting a spreadsheet while working on it.
Possible Workplace Applications
| Scenario | Potential AI Assistance |
|---|---|
| Writing a presentation | Suggest improvements in real time |
| Preparing a strategy document | Identify gaps or inconsistencies |
| Coding | Explain errors visible on screen |
| Data analysis | Help interpret charts or tables |
| Sales calls | Generate summaries and action items |
| Meetings | Track decisions and follow-ups |
| Draft responses using conversation context | |
| Research | Connect information across documents |
The value of such a system would come from eliminating the need to repeatedly explain what the user is doing.
Meeting And Call Recording Could Change Productivity
Another major element of Altman’s comments is the ability to record meetings and calls.
AI meeting transcription is already becoming common across productivity software, but a future ChatGPT could potentially combine meeting information with other personal context.
For example, a meeting discussion could be connected with previous emails, documents and tasks.
Potential Meeting Workflow
MEETING
↓
AI Records Conversation
↓
Transcription
↓
Key Decisions
↓
Action Items
↓
Cross-reference Emails & Documents
↓
Personalised Follow-up
That could make AI considerably more useful as an executive assistant or workplace productivity tool.
However, recording a meeting is different from analysing a document that the user voluntarily uploads. Meetings and calls involve other people, meaning the privacy implications extend beyond the person using ChatGPT.
“Perfect Context” Would Require Much More Data
Altman’s broader vision involves ChatGPT having a highly comprehensive understanding of an individual’s life.
The information required to achieve that would potentially include multiple categories of personal data.
| Data Category | Possible Information |
|---|---|
| Screen activity | Applications, documents and websites |
| Communications | Emails, messages and calls |
| Meetings | Conversations and decisions |
| Files | Documents, spreadsheets and presentations |
| Workflows | Tasks and projects |
| Preferences | Repeated behaviours and choices |
| Historical interactions | Previous AI conversations |
The more sources an AI can access, the more contextually relevant its responses could become.
But the same relationship creates a central trade-off: more context can produce better assistance, while more access creates a larger privacy and security surface.
Privacy Could Become The Biggest Challenge
The proposed capabilities would raise privacy questions far beyond those associated with a conventional chatbot.
A user choosing to share a document with an AI makes a relatively clear decision about what information is being provided. Continuous screen monitoring or call recording would potentially capture information that users did not consciously intend to share at every moment.
That could include confidential business information, financial data, private conversations and information belonging to other people.
Privacy Risk Expands With Data Access
More Data Access
↓
More Context
↓
More Personalised AI
↓
BUT
↓
More Sensitive Information
↓
Larger Security & Privacy Surface
The challenge will therefore be designing controls that allow users to determine exactly what ChatGPT can see, when it can see it and what happens to the resulting information.
Other People Could Be Captured Too
Call and meeting recording introduces a particularly difficult issue.
A user can potentially consent to an AI recording their own activity, but a meeting may include five, 10 or hundreds of other people. Those participants may not have independently agreed to have their conversations processed by an AI system.
The same issue applies to screen sharing and collaborative work.
If a user’s screen contains someone else’s private information, allowing an AI to continuously observe the screen could expose information that the user does not personally own.
This makes consent and workplace policies particularly important for any future system with persistent observation capabilities.
AI Could Become More Proactive
The biggest product change may ultimately be the shift from reactive AI to proactive AI.
Traditional ChatGPT waits for a prompt. A context-aware assistant could potentially identify a situation where assistance would be useful and offer help based on what it sees.
For example, if a user is preparing an important presentation, the AI could potentially identify a weak argument or inconsistent figure before the user asks for help.
Reactive Vs Proactive AI
| Reactive AI | Proactive AI |
|---|---|
| Waits for a prompt | Detects opportunities to help |
| User provides context | AI gathers authorised context |
| Responds to questions | Can suggest actions |
| Works with supplied information | Potentially observes workflows |
| Short interaction | Potentially continuous assistance |
This could fundamentally change how people interact with AI systems.
OpenAI’s Enterprise Opportunity Could Grow
The technology could also have major implications for workplace AI.
Research published in August 2026 using ChatGPT Enterprise data from more than 1,500 organisations and over 17 million messages found that enterprise AI use is expanding across a broad range of knowledge-work tasks, including writing, technical work, communication and information synthesis.
That suggests there is already substantial demand for AI assistance inside organisations.
A system capable of understanding screens, meetings and documents could potentially move AI deeper into everyday workflows.
Enterprise AI Evolution
AI CHATBOT
↓
AI COPILOT
↓
AI CODING AGENT
↓
AI WORKFLOW AGENT
↓
CONTEXT-AWARE AI ASSISTANT
↓
POTENTIALLY
CONTINUOUS DIGITAL ASSISTANT
Each stage increases the amount of work the AI can potentially perform without requiring users to manually provide every piece of context.
The Technology Is Not Yet A Confirmed Product
It is important to distinguish Altman’s prediction from a formal product announcement.
The comments describe what Altman believes could become possible within approximately six months. They do not establish that OpenAI has announced a product called “screen-watching ChatGPT,” nor do they confirm that every capability mentioned will launch on that timeline.
This distinction matters because advanced AI capabilities can face technical, regulatory, privacy and product-design constraints before they become widely available.
What Is Confirmed Vs Speculative
| Area | Status |
|---|---|
| Altman discussed future screen access | Reported |
| Future meeting/call recording capability | Discussed as a possibility |
| Broader personal context | Part of the vision described |
| Six-month timeframe | Altman’s prediction |
| Confirmed consumer launch date | Not announced |
| Exact privacy architecture | Not disclosed |
| Exact data-retention policy | Not disclosed |
The safest interpretation is that OpenAI is pursuing increasingly context-rich AI, but the precise implementation and timeline remain uncertain.
Why This Could Be A Major AI Shift
If implemented successfully, persistent context could make AI significantly more useful.
Users would no longer have to repeatedly tell the system what project they are working on, which documents matter or what was discussed in a previous meeting.
The AI could potentially maintain continuity across tasks.
That could make ChatGPT less like a search box and more like a digital assistant that operates alongside the user throughout the day.
But that same capability could make mistakes more consequential. An AI misunderstanding a single uploaded document is one thing; an AI misunderstanding weeks of personal communications or workplace activity could influence much larger decisions.
The Bigger Picture
Altman’s comments point toward one of the most important directions in AI development: giving assistants much richer context about what users are doing. Screen access, meeting and call recording, emails and documents could allow AI systems to move beyond answering prompts and toward continuously assisting with work and personal tasks.
The trade-off is equally significant. An AI that understands more about a person must have access to more information, creating greater privacy, security and consent challenges. The technology could eventually make AI substantially more useful, but its success will depend as much on user controls and trust as on the underlying model’s intelligence.
Looking Ahead
Over the next six months, the key question will be whether OpenAI can turn Altman’s vision into a practical product while giving users sufficiently granular control over what the AI can access. Screen monitoring and meeting recording would require careful permission systems, clear indicators when recording is active and strong protections for sensitive information belonging to both users and third parties.
If OpenAI eventually delivers these capabilities, the change could represent a major step in the evolution of ChatGPT from a conversational tool into an always-available digital assistant. But Altman’s comments remain a projection rather than a confirmed launch plan, and the final product could look substantially different depending on technical, privacy and regulatory considerations.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
