Science
OpenAI released Astra model and labeled it a Critical capability
It launched with safety triggers; a U.S. trademark for "GPT" was denied.
OpenAI rolled out its ChatGPT Astra model in early September 2026. The launch followed a development schedule that placed the first public access in the week of August 31 through September 6.
The Astra system incorporated built-in safety triggers designed to limit harmful output. These safeguards were described as comparable to those employed by Anthropic's models. The safety features were confirmed as part of the model's specifications on September 26.
On October 1 OpenAI announced that Astra qualified as a "Critical" capability under its internal Preparedness Framework. The classification placed the model among the company's most sensitive AI systems. The move signaled that OpenAI intended to apply heightened oversight to Astra's deployment.
The same day OpenAI's application for a United States trademark on the term "GPT" was rejected. The decision meant the company could not secure exclusive rights to the abbreviation for its products. The denial left the "GPT" label unprotected under U.S. trademark law.
With Astra now publicly accessible and designated as Critical, OpenAI faces both operational responsibilities and branding limitations. The model's safety mechanisms and Critical status shape how developers may integrate it into applications.
Analysis
By embedding triggers similar to those of competing providers, OpenAI demonstrates an effort to align with emerging best practices, which may affect adoption rates among risk-averse organisations.
Classifying Astra as a Critical capability subjects the model to OpenAI's heightened internal controls and could invite additional regulatory focus. Such a designation typically entails stricter monitoring, documentation, and possibly limits on certain high-risk deployments, influencing how the model is packaged for commercial customers.
The denial of a U.S. trademark on "GPT" prevents OpenAI from monopolising a widely recognised brand identifier.
Together, the safety triggers, Critical classification, and trademark rejection shape the operating environment for Astra. OpenAI must balance the model's technical capabilities with compliance obligations and brand positioning, while competitors may seek to capitalize on the open use of the "GPT" label.
What this was built from
Each claim above traces to a market below, at the price it was trading when this edition went to press.