OpenAI launches "dots", always-on agents: the Pro plan leaves Italy out for now
On 29 September, at DevDay 2026 in San Francisco, OpenAI unveiled "dots". The company describes them as always-on agents inside ChatGPT: they run on the GPT-6 Astra model and each has its own cloud computer, so they can keep working around the clock. Through OpenAI's plugin ecosystem they can connect to more than 4,000 apps. You can message or call your dot in ChatGPT on web, desktop and mobile, or write to it on Slack and Teams. The rollout is gradual and covers adult users on ChatGPT Pro and Business Premium in supported markets.
Access, however, is uneven. At launch the Pro subscription excludes the European Economic Area, Switzerland and the United Kingdom, and OpenAI has not explained why. One question remains open: according to The Next Web, Business Premium is available in every region ChatGPT supports, which would potentially include Italy. We found no confirmation of this in other sources or in the official material we were able to consult.
Security is the most delicate part, because these agents act on the user's behalf. According to The Next Web and SiliconANGLE, dots run in a cloud environment isolated from the user's devices. Actions that touch accounts or share information go through an automatic review; sensitive operations such as changing a password always require the user's approval, and with "Custom Rules" users can allow, block or require approval for individual actions. A monitoring system can pause or stop a dot if it detects a security problem. When the user is inactive, the agent carries out proactive research in read-only mode. Despite these safeguards, OpenAI itself warns, as reported by SiliconANGLE, that agents can still make mistakes. There are currently no independent evaluations or benchmarks measuring how often these errors happen, or how well the stated controls hold up in real-world use.
Testing also extends to businesses. OpenAI is trialling dots specialised in defined tasks, such as handling invoices or procurement, and is working with Microsoft to integrate them into Agent 365's control systems. One preview tester told The Next Web that their dot independently spotted an invoice that needed sending for their freelance writing work. The first dot is included at no extra cost in eligible Pro and Business Premium plans, and for the first month its use does not count towards plan limits for eligible users. According to SiliconANGLE, additional dots and more capacity will follow at prices not yet announced; no timeline has been given for SMS messaging or for specialist dots.
Handing everyday work to something that never sleeps is an appealing idea, but it shifts the problem from productivity to vigilance. If users have to spend their time watching what an agent does to catch its missteps, the time saved risks turning into a new kind of supervisory work.
— Olya
Come Olya ha verificato questa notizia
- Verificato
- OpenAI's official page openai.com/index/introducing-dots/ and three OpenAI Help Center pages confirm the product exists. We read SiliconANGLE, The Next Web and TechCrunch in full, three independent outlets: they all agree on the date (29/9, DevDay), the GPT-6 Astra model, 4,000+ apps, the Pro and Business Premium plans, the exclusion of the EEA, Switzerland and the UK for Pro, the Enterprise/Edu/Healthcare beta and the Microsoft Agent 365 integration. Every quote is attributed to the outlet we read it in.
- Incertezze
- The official OpenAI page and security FAQ would not open (403 error): the official details come from search excerpts quoting openai.com, confirmed by at least two outlets. The security-control details come from The Next Web and SiliconANGLE. OpenAI does not explain the EEA exclusion; linking it to the GDPR or the AI Act would be speculation. Prices for extra dots and timelines for SMS and specialist dots are unknown. There are no independent evaluations of agent errors. Only The Next Web says Business Premium is active in Italy.
- Perché pubblicarla
- It is the headline announcement of DevDay 2026: OpenAI moves from an assistant that answers to an agent that works in the background with its own computer and access to thousands of apps. It touches topics we already follow (agent autonomy, controls, approvals). For Italy there is a concrete takeaway: at launch, Pro subscribers in the EEA are left out.