OpenAI CEO Sam Altman has offered a striking glimpse into the company’s vision for the next generation of AI assistants, suggesting that a descendant of ChatGPT could, within six months, continuously understand what users are doing across their screens, meetings and calls and build a much deeper picture of their lives.
Altman’s comments, which have circulated widely online, point to a future in which AI moves beyond the traditional chatbot model and becomes a persistent digital assistant with access to a user’s broader digital context. According to reports quoting his remarks, such a system could watch a user’s screen, record meetings and calls and maintain a near-complete understanding of the context surrounding the user’s activities.
Importantly, Altman was describing a future capability rather than announcing a product launch or confirming that such a system will be released within six months. The remarks nevertheless offer an indication of the direction in which OpenAI sees personal AI assistants evolving.
The shift could fundamentally change the value proposition of AI assistants. Instead of relying primarily on information users deliberately type into a chatbot, an always-on assistant could potentially derive context from emails, documents, meetings, calls, applications and activity on a computer.
That raises a bigger question: if an AI system accumulates years of a person’s conversations, work, preferences and decisions, could that information become one of the most powerful forms of competitive advantage for the company operating the system?
The issue is particularly significant because AI companies are increasingly competing to make their assistants more personalised and proactive. OpenAI has been exploring increasingly autonomous AI systems, while recent reporting has highlighted work on persistent AI agents capable of continuing tasks and acting proactively based on previous interactions and user knowledge.
Microsoft’s Recall experiment offers a warning
Microsoft has already demonstrated both the potential and the risks of this approach with Windows Recall.
Recall was designed to periodically capture snapshots of activity on a user’s computer so that the information could later be searched using natural-language queries. Security researchers raised serious concerns about how the captured information was stored and how easily it could potentially be accessed.
Microsoft subsequently delayed the public rollout amid privacy and security concerns. The company later redesigned Recall’s security architecture, including additional protections around the snapshots and associated data.
The experience highlighted a central problem with persistent AI: the more useful the system becomes, the more sensitive information it potentially needs to access.
An assistant that can remember a user’s previous conversations may be useful. One that can understand everything happening on a user’s computer, hear meetings and calls and connect that information with emails and documents would be considerably more powerful — but would also create a far larger security and privacy risk.
There is also a less obvious business implication.
If an AI assistant becomes the repository of years of personal and professional context, switching to a competing service could become increasingly difficult. Users might not simply be changing applications; they could effectively be leaving behind an accumulated digital history that makes the incumbent assistant useful.
That could turn memory and context into a form of competitive moat.
The idea is still speculative. There is no evidence that OpenAI has announced a product designed to record every call or continuously monitor users, nor that it intends to make such functionality mandatory. Altman’s comments should therefore be read as a projection of where AI assistants could be heading, rather than confirmation of a forthcoming product.
But the direction is clear: the next battle in AI may not be over who has the smartest chatbot. It could be over who becomes the system that knows the most about you.
And that leaves consumers with a question that technology companies will increasingly have to answer: How much of your digital life are you willing to place in the hands of one AI company?
