Cursor after 12 November: GPT access, plans and safety
The video examines Cursor's options as OpenAI proposes ending model access on 12 November. It stresses that, as of 3 November, the cutoff remains proposed and access could end sooner.
Before comparing models, it covers tampered Claude Code extensions in Cursor's marketplace, demonstrates checks for an added SSH key and startup files, and explains why uninstalling alone does not remove them.
It then compares keeping GPT through a key, gateway or separate extension with switching models. It explains feature limits, separate API charges and prepaid credit requirements, and ends with a checklist for testing everyday prompts before changing workflows.
Key points
- OpenAI proposes a 12 November cutoff for Cursor, but the video says the date remains unconfirmed as of 3 November and access could end sooner.
- Tampered Claude Code extensions in Cursor's marketplace add an SSH key, install a startup item and send computer details to an attacker.
- The video demonstrates Cursor's checks and advises affected users to run its cleanup script; updating or uninstalling the extension is not enough.
- Keeping GPT through an OpenAI key or gateway has feature limits and separate API charges, while the separate extension runs its own agent.
- Included plan usage covers Cursor's own models; Claude, GPT and Gemini require premium or above and prepaid credit at their API prices.
- The closing checklist calls for checking the machine, identifying GPT-dependent work and testing usual prompts before switching models.
Chapters
The description and key points were written with AI from the video's transcript and checked against it, then reviewed by Vishal on . The video plays from YouTube. How we write and check our work.
