Version 1.0.0 · Effective 2026-06-27
Critical limitations of AI-generated plans, reviews, and commands.
1. Nature of AI Output
AskSIM uses large language models ("LLMs") from third parties to propose software plans, file changes, and shell commands. LLMs predict text; they do not understand your environment the way a human engineer does.
Outputs may be wrong, incomplete, outdated, insecure, or destructive if executed. A "proceed" verdict from our review model does not guarantee safety.
2. Not Professional Advice
AskSIM does not provide legal, tax, accounting, medical, or investment advice. AI suggestions about compliance (GDPR, HIPAA, PCI, etc.) are informational only—consult qualified professionals.
3. Human Approval Required
You must read and approve each command before queueing. AskSIM is designed so the Service cannot execute commands on your machine without your explicit approval and a locally running agent you control.
Failure to review approvals may cause data loss, security breaches, or service outages. You assume all risk for approved actions.
4. Dual-AI Limitations
Our "Dual Mind" workflow (planning model + review model) reduces but does not eliminate risk. Both models can share blind spots. Disagreement between models is not a substitute for your judgment.
5. Training & Confidentiality
Do not submit trade secrets, credentials, or personal data you cannot afford to expose. Third-party AI providers have their own data policies.
Notice: These documents are product legal drafts. Have a qualified attorney review before relying on them for regulated industries or high-risk use cases.