We've shipped a major accuracy upgrade to the FlySafe risk engine.
Sharper scores. We re-engineered how the score composes evidence: a FIR's score now reflects the current weighted evidence — exactly. Regions with chronic background indicators but no present airspace impact no longer read as elevated; the score responds to what is happening now, not to what a region's history suggests.
New evidence floor. A FIR with a confirmed critical NOTAM in force can no longer score below 55 ("high"), even when every other signal is quiet — hard regulatory evidence now sets a minimum.
Versioned. Every score now carries a scoring_version field (sv2). Historical scores are unchanged (sv1) — we don't rewrite history.
What you'll see: a one-time step change in 30/60-day sparklines on July 16, and noticeably fewer FIRs at high/critical. Validated against our event ground truth: significantly fewer false alerts at every threshold, and the missed-event rate held below 1%.
Programmatic API Access
The detection behind this story runs continuously. For operational decisions requiring per-FIR, sub-hour resolution, programmatic API access is available to verified operators.
Request API access
Per-FIR risk · live alerts · 300+ FIRs
Information is accurate as of the publication date. FlySafe uses exclusively publicly available data.
← Roadmap & changelog