Checking status…
Real-time status of every public service. If we're down and this page is down too, try the external mirror at cronitor.
Recent incidents
No incidents in the last 90 days.
Uptime history will appear here once we have 30 days of polling data. Open incidents are also posted to @deepaidetector.
Subscribe to status updates
Get an email when an incident opens or resolves. We do not send marketing on this list — incidents only. Unsubscribe at any time from the footer of any message.
Subscribe via emailAbout the Deep AI Detector status page
This page reports the live operational state of every public Deep AI Detector service that end users and API integrations touch. Components include the AI content detection pipeline, multilingual translation, database, result cache, rate-limit store, corpus storage, and exports storage. The data refreshes every 30 seconds and is mirrored to an independent external monitor so you have a fallback view if the primary status page is itself affected by an incident.
How the checks work
Each component runs a deep readiness probe that exercises the real dependency — not just a TCP ping. The database check writes and reads a sentinel row. The cache check writes, reads, and deletes a test key. The ML inference check sends a known sample to the detection model and verifies the response shape. This means a degraded component is flagged before users notice — and conversely, "operational" means the dependency is actually serving traffic, not just answering pings.
Service levels
Deep AI Detector targets 99.9% monthly availability on the AI detection endpoints and 99.95% on the public website. Paid Enterprise plans include a contractual SLA with service credits for downtime that exceeds the published target. The pricing page outlines the tiers; sales can quote a custom SLA for institutions with stricter requirements.
Incident communication
When an incident opens, the banner above flips from green to yellow (degraded performance) or red (active outage). Subscribers to the incident mailing list receive a notification within minutes, and a public post-mortem is published within 72 hours of any incident lasting more than 30 minutes. We will not publish a post-mortem for short transient blips — the bar is "did this materially affect customers."