OpenAI Accelerates Work on GPT-6 With Focus on Memory and Personalization

OpenAI Chief Executive Sam Altman disclosed last week in San Francisco that work on GPT-6 began almost immediately after the release of GPT-5, offering a rare look into the company’s product pipeline and its growing challenges.

Although no launch date has been set, Altman suggested GPT-6 will arrive more quickly than the gap between the two previous versions. Unlike earlier models, the upcoming system is designed not only to answer questions but also to adapt to individual users. By incorporating memory, ChatGPT could retain information about personal habits, preferences, and behaviors, enabling people to create tailored chatbot experiences. “People want memory,” Altman remarked, describing personalization as essential to the next stage of development.

Altman also revealed that OpenAI has collaborated with psychologists to study emotions and track levels of happiness, research that has guided the evolution of the product. While this data has not yet been shared publicly, he indicated it may eventually be released.

Future iterations will also align with a recent executive order from the Trump administration requiring AI systems used by federal agencies to remain ideologically neutral and customizable. Altman emphasized that the model should operate from a centrist baseline, allowing users to adjust settings according to their own preferences.

The remarks follow a difficult rollout of GPT-5, which faced criticism for feeling less engaging and less effective than GPT-4. Altman acknowledged that the launch could have been handled better but noted improvements have since been introduced, including updates that make GPT-5 “much warmer” in tone. Even so, concerns remain around memory features, particularly because temporary data storage is currently unencrypted. He confirmed encryption is planned, though no timeline has been set, stressing the importance of robust privacy protections for sensitive areas such as health and legal queries.

Altman added that the company is exploring adjacent technologies, such as brain-computer interfaces that could allow users to think a command and receive a ChatGPT response. He also pointed to interests in robotics, energy, novel computing substrates, and faster data center infrastructure.

For now, ChatGPT remains the company’s central product, with Altman saying he relies on it daily for both professional and personal use. Yet he admitted that conversational AI is approaching a natural limit. “The models have saturated the chat use case,” he said. “They’re not going to get much better… and they’re probably going to get worse.” His comments suggest OpenAI is in a pivotal phase—managing criticism of GPT-5 while pushing toward GPT-6, which promises to bring major advances in memory and personalization.