Composer 1 LLM security

Support Help
1

Good day,
Is Composer 1 LLM model derived from Chinese llm model(s)?

If so, is it totally secure/safe, even for USA enterprise use? I mean in regards to if Chinese llms try to inject any vulnerabilities when they produce code or similar… I realize our code does not get shared if we are opted out in our Cursor account portal but can LLM produce negative code in light of hacking/compromising security if based on Chinese model?

Thanks for your time,
Shawn

2

Hey, thanks for the question.

All models in Cursor are hosted on infrastructure in the US, Canada, and Iceland - either by the model provider, a trusted partner, or Cursor itself.

For enterprise usage in the US, Cursor provides:

  • SOC 2 Type II certification
  • GDPR compliance
  • Privacy Mode with zero data retention
  • Detailed info on the Trust Center: https://trust.cursor.com

You can find detailed security information here: Security · Cursor

If you need specific provenance details for Composer models for an enterprise security assessment, please contact sales via Enterprise · Cursor - they can share detailed technical docs for your security review.

3

Thank you for your reply, Dean.
You mentioned the various security features for enterprise usage in US; Does that mean it is not as secure on individual plans?
Also can you comment on my particular question of can Composer 1 model produce negative code in light of hacking/compromising security if based on Chinese model? In other words, I know AI makes mistakes, of course… but by “safe” I mean, if Composer 1 model has one of the Chinese models as its base, there is the concern of what if the Chinese models try to intentionally but perhaps implicitly make code at times that is exploitable/hackable?