> For the complete documentation index, see [llms.txt](https://ggai.gitbook.io/ggai-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ggai.gitbook.io/ggai-docs/security-and-abuse-prevention.md).

# Security and Abuse Prevention

The paradox of building a “chaos tool” is that it can itself be abused if not carefully controlled. Key security considerations Glitch Gremlin takes:

1. **Multi-sig Control:** The upgrade authority for the program and mint is locked behind a multi-sig. This prevents unilateral changes or malicious re-deployments.
2. **Request Throttling:** We rate limit how many chaos requests can be initiated within a certain time to avoid spam or overloading the system.
3. **KYC or Access Controls (*****Coming Soon*****):** In advanced scenarios, the team may require certain verifications for large-scale or potentially destructive chaos tests, especially if they risk collaterally damaging public networks.
4. **Verified Off-Chain Engine:** The AI engine signs finalization transactions with a known, verifiable keypair. This ensures that only the official chaos simulations can mark requests as completed.
5. **Third-Party Audits:** *Coming Soon*