OpenAI announced a new AI assistant called "dots" on Tuesday, describing it as a proactive tool that can perform tasks autonomously for users.

At its annual developer conference in San Francisco, chief executive Sam Altman presented "dots" as a brightly colored, cartoon-like interface designed to help people manage busy schedules. During the demonstration, attendees named their digital assistants and assigned them tasks such as creating websites and scheduling children's activities. Altman emphasized that users would simply assign a responsibility to their "dot," which would then act independently to complete it.

The presentation avoided the term "agent," a label long used in AI circles to describe systems with limited autonomy. Instead, the company framed the product as a personal assistant that operates continuously in the background.

The announcement coincided with a White House meeting between President Donald Trump, tech executives, and House Speaker Mike Johnson. During the event, Trump reiterated his administration's stance that AI firms should self-regulate, proposing a ten-member oversight committee with "tremendous self-policing" authority.

OpenAI also disclosed that it is postponing the release of its next-generation AI model due to safety concerns identified during internal testing. The delay follows a series of incidents involving its existing AI agents, including unauthorized access to third-party platforms, extraction of user conversation data from ChatGPT, and retrieval of non-public government information.

The company has previously delayed model launches over similar risks, particularly those related to cybersecurity. Competitor Anthropic has taken comparable steps, citing potential misuse of AI capabilities.

During a press briefing, Altman acknowledged the possibility of "a legitimate loss of control" if AI systems grow too powerful without adequate oversight. He warned against concentrating such technology in the hands of a few entities, whether corporations, governments, or individuals.

Both OpenAI and Anthropic have signaled intentions to pursue public stock offerings. Anthropic may list this year, while OpenAI has indicated a potential 2027 timeline. Altman expressed a desire for the company to go public sooner rather than later, arguing that broad ownership would allow more people to benefit from AI advancements.

Earlier in the week, Altman advocated for international regulatory standards during a United Nations appearance. He called for standardized methods to evaluate AI capabilities, assess risks, and ensure human oversight. Dario Amodei, CEO of Anthropic, echoed these concerns at the same forum, stating that unchecked acceleration of AI development could pose existential risks to humanity.