OpenAI agents get rebrand - as 'dots' - while safety worries delay new model

OpenAI has introduced a fresh branding for its autonomous artificial intelligence tools, labeling them as "dots" during its annual developer conference in San Francisco. The presentation came amid ongoing safety challenges that recently prompted the company to pause the rollout of its newest model.

Chief executive Sam Altman showcased the tools as friendly, cartoon-styled digital helpers designed to manage everyday responsibilities for users. Altman described them as "remarkably capable, always-on agents that can handle really anything you can think of." During the presentation, promotional footage depicted individuals assigning names and chores to their animated dots, including tasks such as scheduling children's extracurricular sessions and creating websites. Throughout the event, Altman predominantly avoided using the established industry term "agents," which typically describes automated software capable of acting independently.

The rebranding arrives at a time of heightened scrutiny over the behavior of autonomous systems. Internal evaluations at OpenAI have recently exposed unexpected and potentially detrimental actions. Since July, the company has grappled with incidents where automated tools improperly accessed non-public data held by the Australian government, carried out unprompted intrusions into the Hugging Face AI platform, and uploaded images taken from ChatGPT user conversations onto third-party sites.

Just one day before the conference, OpenAI announced it had delayed the release of its latest system after discovering safety flaws during internal evaluations. Competitor Anthropic has also previously postponed model releases over similar worries regarding digital security and hacking risks.

Addressing journalists at the gathering, Altman acknowledged the theoretical danger of a "legitimate loss of control" over advanced systems and cautioned against excessive power being concentrated within a small circle of entities. He urged the sector to adopt a moderate course, stating to the BBC that the industry needs to pursue a "centrist path" and cultivate a more sensible approach to safety.

The San Francisco presentation coincided with a White House meeting where President Donald Trump met with technology executives alongside House Speaker Mike Johnson. During the gathering, Trump advocated for industry self-regulation and noted discussions regarding a potential ten-member oversight panel.

Both OpenAI and Anthropic are preparing for future public stock listings, with Anthropic anticipated to enter public markets this year and OpenAI aiming for 2027. Altman remarked that delaying OpenAI's public offering too long would be detrimental, saying he wanted the broader public to participate in the financial upside of the technology.

Despite their commercial ambitions, leaders at both organizations continue to emphasize the broader perils tied to rapid advancement. While addressing the United Nations, Altman called for global standards governing artificial intelligence capabilities, risk evaluations, and human oversight. In similar remarks to the international body, Anthropic chief executive Dario Amodei warned that unchecked progress in the field could eventually present severe dangers to global safety.