OpenAI launches dots: AI agents working 24 hours for you (and the security paradox)

OpenAI presented the "dots", personal AI agents capable of working autonomously on the user's goals 24 hours a day. The announcement, reported by Al Jazeera, describes an assistant "made to deal with everything" — and arrives a few weeks after CEO Sam Altman publicly defends a slowdown in the development of more advanced AI and the company itself filing a model considered misleading for security reasons.

What are dots and why aren't they just another IA 24 chatbot? h

The central difference is in the operating model. A chatbot responds to a request and stops. The dots, according to the launch description, receive a goal and continue working on it in the background, continuously, without requiring a new command at each stage. It is a change of category: exits the "responsive assistant", enters the "executing assistant".

OpenAI did not detail at source which specific tasks dots already complete alone on launch day, nor which external services are connected to them. What the company communicates is the proposal for broad coverage — hence the slogan of an assistant who handles everything.

Why the timing of 24h AI agents exposes a paradox

This is the most delicate point in the announcement. Weeks earlier, Altman defended the deceleration of border AI development — a speech that puts the CEO in line with requests for regulatory caution. In the same period, OpenAI filed a model that was considered misleading for security reasons.

Now the company launches a product whose premise is prolonged autonomy: agents acting on their own, for long periods, without constant supervision. The tension between the two movements is evident and should dominate the public debate in the coming weeks.

How 24h AI agents stand up to competitors

OpenAI does not enter an empty field. The market for autonomous agents already has established names, and some of them use OpenAI's own technology under the hood — such as the Amazon Bedrock Managed Agents, which put previously managed agents in AWS. The Meta launched the Muse Spark 1.1, focused specifically on autonomous agents, while Mistral and Anthropic attacked the same problem through different paths — open models and corporate agents with more aggressive price.

The differential that OpenAI tries to sell with dots is not the isolated capacity of the model, but the proposal of a personal agent who accompanies the user throughout the day, rather than a corporate workflow tool.

What changes for those who develop or hire 24h AI agents in Brazil

Agents running 24 hours change the account three ways for Brazilian teams. First, the cost: continuous inference consumption is unpredictable, and cloud budgets were designed for peaks of use, not for permanent operation. Secondly, the audit: when the agent makes a decision alone at 3:00 in the morning, someone needs to be able to rebuild what happened and why. Third, responsibility: automated decisions that affect people require traceability, and this discussion remains open in the country.

For those who hire, the practical question ceases to be "which model is better" and becomes "how many agents I can actually supervise".

What's the announcement from the 24-hour AI agents still not responding

They were left out of the initial disclosure essential information: price, country availability, supported languages, limits of autonomy and what mechanisms of human supervision exist when a dot err. Nor is it clear how the product relates to the model filed for security reasons, nor do dots inherit any of the restrictions applied to it. Without this data, the "made to deal with everything" remains, for now, a promise of positioning.

Frequently Asked Questions about AI Agents 24 h

Dots work without any human supervision?

The proposal is autonomous operation 24 hours a day in the goals defined by the user, but OpenAI has not detailed in the source which mechanisms of supervision or interruption exist.

How much does it cost to use dots?

OpenAI did not disclose price or collection model in the ad reported by Al Jazeera. Without public values, any estimate would be speculation.

Are dots already available in Brazil?

There is no confirmation of availability by country at source. The company also did not inform whether the launch is global from day one or restricted to some markets.