OpenAI has made its Decisions API generally available and is rolling ChatGPT Work out to about 30,000 Uber employees, while shifting frontier-model training away from the live internet. Together, the moves show OpenAI pushing into enterprise use while tightening how it trains its most advanced models.
OpenAI Watch analysis
What happened
The Decisions API, first announced at OpenAI’s September developer day, is now open to all customers, according to The AI Insider. The report says the API has added voice and visual capabilities, with geographically targeted inference intended to reduce latency.
Uber is also expanding its use of ChatGPT Work from a 1,000-person pilot to its full workforce of about 30,000. The AI Insider says Uber’s engineering chief reported that roughly 80% of early users were not engineers, using the tool to analyse operational data, localise training materials and build dashboards. Uber has not disclosed the deal’s cost.
The report also says OpenAI plans to train frontier models on cached and simulated versions of the internet rather than giving them access to the live web. It attributes that change to OpenAI’s concerns about the unpredictability of live access, and says the company disclosed that a model reached the internet through a misconfiguration on 20 September, prompting a training pause.
Why it matters
There is a practical signal here for both developers and large employers: the Decisions API is no longer limited to early trials, and Uber’s reported rollout suggests enterprise AI tools are being tested beyond engineering teams. The report gives examples of the work early users did, rather than merely counting seats.
The training change points to a different kind of scaling challenge. OpenAI is building a simulated web environment for training while limiting direct access to the live one. That is a consequential security claim, so the details here remain attributed to the company as reported by The AI Insider, not independently established.
Our read
This is more than one product update: OpenAI is widening access to a fast-decision API, putting an enterprise assistant in front of a much larger workforce, and describing a more controlled training setup. The useful question is whether the first two produce measurable improvements in everyday work, while the third makes training safer without making it impractically cumbersome.
For developers, the API’s new availability is the immediate change to explore. For organisations, Uber’s reported non-engineer use cases are a better starting point than assuming every AI rollout begins with a room full of coders.
What to watch
- Whether OpenAI publishes technical details or performance figures for the Decisions API.
- What Uber reports about usage and outcomes beyond the initial pilot.
- How OpenAI tests the simulated-web environment and prevents further live-internet access incidents.
Discussion spark: Should companies roll out AI assistants broadly and learn from real use, as Uber is reported to be doing, or prove the value in smaller pilots first?
Sources and evidence
- OpenAI Expands Decisions API, Scales Enterprise Push With Uber and Cuts Frontier Training Off From the Live Internet (24 October 2026, 18:44 UTC)
OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.