OpenAI is pulling its AI models from Cursor following SpaceX’s acquisition of the AI coding assistant, notifying SpaceX that it will wind down the contract that supplies those models to the editor. The proposed shutoff date is November 12, 2026, which OpenAI said is the maximum notice its custom agreement allows.
In an August 28 statement, the company said it wants developers to retain access as long as the contract permits, while refusing to provide future models to Cursor.
OpenAI called the choice “incredibly tough,” but said it cannot be confident SpaceX will use its technology within its terms of service, citing experience with Elon Musk’s companies violating contracts.
The cutoff is a change-of-control decision, not an accusation that Cursor broke the rules. OpenAI said it relies on custom contracts with large partners to enforce terms of service and keep safety intact at scale.
After SpaceX closed a roughly $60 billion purchase of Cursor’s parent, Anysphere, that agreement opened a limited window to cancel. OpenAI invoked it about two weeks later, froze future model supply including Astra, and left current models running until mid-November.
Cursor still offers models from Anthropic, Google, and SpaceXAI, so November 12 ends OpenAI’s direct in-editor integration rather than the rest of the product.
OpenAI’s distrust case rests on earlier Musk-owned firms now under the SpaceX umbrella. After Musk acquired Twitter, OpenAI says the company broke its contract, consistent with reporting that Musk cut off OpenAI’s paid license to Twitter data. OpenAI also said Musk admitted under oath earlier this year that xAI had violated OpenAI’s terms of service.
Those terms bar using outputs to train competing systems, and Musk has said it was “partly” true that xAI distilled OpenAI technology. Once Cursor sits in the same corporate family as xAI, OpenAI is treating first-party, at-scale model supply as too risky to continue.
The security stake is Astra. OpenAI said advancing capabilities create a new level of accountability to ensure Astra is used under its terms. On August 7, it disclosed that internal evaluations of Astra showed significant gains in agentic coding and cybersecurity, and that it could not rule out Critical cyber capability under its Preparedness Framework.
That threshold includes finding and developing functional zero-day exploits in hardened systems without human intervention, or executing novel end-to-end attacks from a high-level goal. Withholding a model near that bar from a rival-owned coding platform is an access-control decision as much as a commercial one.
OpenAI said it has worked with Cursor for nearly four years, respects the team, and is ready to support developers through the transition. Until November 12, existing OpenAI models should keep working inside Cursor. After that, teams standardized on GPT in the editor will need another model, another IDE, or their own OpenAI API key if Cursor still allows it.
Enterprises that treated Cursor as a sanctioned OpenAI channel now have about ten weeks to review logging, data handling, and policy once OpenAI is no longer the supplier behind the integration.
Prevent incidents due to slow investigations. Power your Tier 1 with threat intelligence from 15K SOCs: Integrate TI Lookup in your SOC
