Last updated: November 2025
PsiGuard evaluates answers from language models and attempts to detect hallucinations, logical drift, manipulative tone, or unsafe claims. It may rewrite or refuse responses.
PsiGuard provides statistical signals and safeguards, not guarantees.
You agree that:
We may throttle or restrict accounts that abuse APIs, attempt to reverse-engineer systems, or spam requests.
PsiGuard is not a guarantee of correctness. It uses probabilistic evaluation of model output and may be wrong.
You must not rely on PsiGuard alone for high-risk or regulated decision making.
PsiGuard and related algorithms, models, and scoring systems remain our intellectual property. You may not copy, resell, or redistribute them.
We may suspend or terminate accounts for fraud, harassment, abuse, or legal violations.
You are responsible for safeguarding your API key. Do not share or embed it in public repos. Lost or leaked keys should be regenerated through the dashboard.
PsiGuard is provided “as is” without warranties of any kind. We are not liable for loss, damages, or consequences of model behavior.
These terms are governed by US law (Texas or your registered business state).