1. Third-Party Destinations
Short links lead to websites we don't own or control. We aren't responsible for their content, availability, privacy practices or security. Open links only from people you trust.
2. Safety Verdicts Are Not Guarantees
The Guardian combines rule-based checks with AI models. A “Looks Safe” verdict means our checks found no warning signs at the time of checking; it doesn't certify that a site is safe. Pages can change after they're checked, and AI models can make mistakes. A warning may also appear on a legitimate site.
3. AI-Generated Content
Summaries come from each page's own title and description, which the site owner controls and may be inaccurate. Alias suggestions are ranked automatically. Risk decisions from Jev by TypeSafe are probabilistic, and we only act on them above a confidence threshold. Review AI output before relying on it.
4. Analytics Are Estimates
Unique visitor counts use a HyperLogLog estimate with about 1.6% standard error. Bots are filtered using their user agent, which isn't perfect. Treat numbers as directional, not exact.
5. Demonstration Project
Comet is built to demonstrate backend and system design. It's run with care, but without service-level guarantees. Don't rely on it for critical or long-lived links.
6. Reporting Abuse
If a short link leads somewhere harmful, email hello@example.com with the link and we'll review it.