OpenAI Introduces AI Agents Dots and Stops GPT 6.1 Release

OpenAI invited to the developer conference in San Francisco. However, new developments by the AI ​​company are overshadowed by the debate about security risks and competition from Muse.

Open AI CEO Sam Altman speaks at the OpenAI Developer Day: The new model is intended to be an even more personal assistant in everyday life. Photo: Getty Images via AFP

San Francisco, Munich. OpenAI has introduced a constantly active AI agent that will independently take on tasks for users. This puts the company up against competitors like Meta Platforms in a growing market for AI-powered digital assistants.

The service is called “Dots”. According to OpenAI, he can operate a computer and pull information from various linked applications to research, design documents and develop software. The company announced this on Tuesday at the OpenAI Developer Day.

Users control the agent via messages or calls in ChatGPT or write to them via Slack or Microsoft Teams. In the future, this will also be possible via SMS, OpenAI announced. The company's own engineers are already using the virtual assistant to fix dozens of software errors every day.

OpenAI CEO Sam Altman unveiled the new tool at the company's developer day on Tuesday - along with an updated version of the Sol model and a shared workspace for teams working with AI.

At the same time, OpenAI is restructuring its subscription offering: a new $500 Premium subscription offers higher usage limits and faster processing, while certain usage limits are lowered in the $200 plan.

Discussions with the US government are likely to be ongoing parallel to the event

“It’s like an AI helper that always has your back,” Altman said. He emphasized the security aspects of the tool: users could specify what the assistant was allowed to access. “So you can give your Dot as much responsibility as you want,” he said.

The company is also testing specialized dots for corporate customers, for example for email marketing, accounting and legal analysis. “We want people to discover what’s possible with AI – even things they never imagined,” said Altman.

The ChatGPT maker wants to increase sales and usage of its more advanced AI tools while facing increased scrutiny over security concerns about its technology. In parallel to the event, leading AI managers, including OpenAI President Greg Brockman, were scheduled to attend a lunch with US President Donald Trump to discuss AI risks - after public concern about possible dangers from artificial intelligence had recently increased.

Sam Altman (l) and US President Trump: The OpenAI boss sat at the table with Trump at the last G7 summit in Evian. Photo: AFP

A day before the conference, OpenAI announced that it would not be presenting the planned version GPT-6.1 of its top model Astra. The company referred to increased risks after there had recently been several incidents involving AI systems acting on their own initiative during hacker attacks.

Technology

OpenAI stops publishing new AI model

The new GPT-6.1 Sol model announced on Tuesday is described by OpenAI as “close to Astra.” The company also apologized to Australia this week after some of its AI models breached Australian government websites.

Dots runs on Astra's existing GPT-6 version and joins an increasingly crowded market for AI agents - including offerings from Meta, SpaceX and younger start-ups such as Instinct.

The Muse service in particular has recently gained traction among consumers, for example by taking on tasks such as online shopping or automatically entering important appointments from emails into a family calendar.

OpenAI wants to be one step ahead of the competition

Technology companies have been trying to develop convincing personal AI assistants for years - with varying degrees of success. Today's most advanced chatbots already fulfill this function in many ways, particularly through increasing capabilities in complex tasks. Companies are now relying on more intuitive interfaces and connections to external apps to bring noticeable progress.

Like Meta, OpenAI also has a large audience that can accelerate the spread of new services. On Tuesday, the company announced that ChatGPT is now used by more than 1.2 billion people weekly - up from 1 billion in the summer.

OpenAI wants to activate Dots for paying users from Tuesday. First, each user gets access to a single “Dot” as an assistant. In the long term, the company expects users to manage a whole range of them.

IPOs of AI companies are proving difficult

OpenAI has long been in an intense competition with long-time rival Anthropic for corporate customers. Both companies have filed confidential documents for an IPO. Anthropic could go public as early as this fall.

Related topics

OpenAISam AltmanChatGPTDonald TrumpAnthropicArtificial Intelligence

OpenAI says it no longer expects to go public in 2026 because the company initially wants to focus on security concerns with its technology. OpenAI was considered to be on the defensive for much of the year, but has since focused its product offering and seen a new push in tools that, among other things, simplify software development.

OpenAI's annualized revenue rate accelerated over the summer, surpassing $40 billion, Bloomberg reported in August.

With agency material.

More: AI researcher resigns: “They are playing with our lives”

Published according to the editorial standards of the Handelsblatt. You can find more information in our guidelines.

More on the topic of our partners display

remind.me Take advantage of current low electricity/gas prices before prices rise again