OpenAI Pulls Models from Cursor After SpaceX Acquisition

Key Takeaways

• OpenAI is discontinuing its AI model supply to Cursor, an AI coding assistant, following Cursor’s acquisition by SpaceX.
• The decision, effective November 12, 2026, stems from OpenAI’s concerns about potential terms of service violations by Elon Musk’s companies.
• This move impacts developers currently relying on OpenAI models within the Cursor editor, who will need to transition to alternative AI models or development environments.
• The withdrawal includes future access to advanced models like Astra, which OpenAI has identified as having “Critical cyber capabilities.”

OpenAI has announced it will cease providing its artificial intelligence models to Cursor, an AI-powered coding assistant, in response to its recent acquisition by SpaceX. OpenAI formally notified SpaceX of its intent to terminate the contract supplying these models, with a proposed discontinuation date of November 12, 2026. This timeframe represents the maximum notice period permitted under their custom agreement.

In a statement released on August 28, OpenAI clarified its commitment to allowing developers continued access to current models for the duration of the contract, but explicitly stated it would not furnish future models to Cursor. The company described this choice as “incredibly tough,” yet necessary due to a lack of confidence that SpaceX would adhere to its terms of service, citing prior experiences with Elon Musk’s other ventures allegedly breaching contractual obligations.

This termination is a consequence of a change-of-control clause rather than an accusation of wrongdoing by Cursor itself. OpenAI emphasized its reliance on tailored contracts with significant partners to uphold its terms of service and maintain safety protocols at scale.

The agreement between OpenAI and Anysphere, Cursor’s parent company, included a limited window for cancellation following SpaceX’s approximate $60 billion acquisition. OpenAI exercised this right roughly two weeks after the acquisition, immediately halting the supply of future models, including the advanced Astra, while allowing existing models to function until the mid-November 2026 cutoff.

It is important to note that Cursor will continue to offer models from other providers, including Anthropic, Google, and SpaceXAI. Therefore, the November 12, 2026, deadline marks the end of OpenAI’s direct integration within the editor, not the complete functionality of the product.

OpenAI’s Rationale for Distrust

OpenAI’s decision to withdraw services is rooted in what it describes as a pattern of non-compliance by Musk-owned entities now operating under the SpaceX umbrella. Specifically, OpenAI alleges that after Musk acquired Twitter, the company violated its contract. This aligns with reports indicating Musk cut off OpenAI’s paid license to Twitter data. Furthermore, OpenAI stated that earlier this year, Musk admitted under oath that xAI had infringed upon OpenAI’s terms of service.

These terms explicitly prohibit using OpenAI’s outputs to train competing systems. Musk himself has acknowledged that it was “partly” true that xAI had distilled OpenAI technology. With Cursor now part of the same corporate family as xAI, OpenAI views the continued provision of its models, particularly at scale, as an unacceptable risk.

The Security Implications of Astra

A significant factor in OpenAI’s decision is the security implications surrounding its advanced Astra model. OpenAI highlighted that the increasing capabilities of such models necessitate a heightened level of accountability to ensure their usage aligns with terms of service. On August 7, OpenAI revealed internal evaluations of Astra demonstrated substantial advancements in agentic coding and cybersecurity. Crucially, it could not definitively rule out Astra achieving “Critical cyber capability” under its Preparedness Framework.

This “Critical cyber capability” threshold includes the ability to independently discover and develop functional zero-day exploits in hardened systems, or to execute novel, end-to-end attacks from a high-level objective without human intervention. Consequently, withholding a model nearing this sophisticated capability from a coding platform owned by a competitor is as much an access-control decision as it is a commercial one.

OpenAI expressed its long-standing collaborative relationship with Cursor, spanning nearly four years, and conveyed respect for the Cursor team. The company affirmed its readiness to support developers through this transition period. Until November 12, 2026, existing OpenAI models should continue to function within Cursor. After this date, development teams that have standardized on GPT within the editor will need to migrate to an alternative AI model, a different Integrated Development Environment (IDE), or potentially use their own OpenAI API key if Cursor continues to support such integration.

Enterprises that previously considered Cursor a sanctioned channel for OpenAI technology now have approximately ten weeks to reassess their logging practices, data handling protocols, and internal policies, given that OpenAI will no longer be the underlying supplier for this integration.

What You Should Do

• Evaluate Alternatives: Developers and enterprises currently using OpenAI models within Cursor should begin evaluating alternative AI coding assistants or integrating other AI models (e.g., Anthropic, Google, SpaceXAI) supported by Cursor.
• Review API Key Usage: If your team relies on direct OpenAI API keys within Cursor, confirm if this functionality will persist post-November 2026 and assess its viability as a long-term solution.
• Update Security Policies: Enterprises should review and update their data handling, logging, and security policies to reflect the change in AI model supplier, ensuring continued compliance and data integrity.
• Plan for Transition: Establish a clear migration plan and timeline to ensure a smooth transition away from OpenAI models in Cursor before the November 12, 2026, deadline, minimizing disruption to development workflows.

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.