Two armed suspects carjacked a 21-year-old at gunpoint on the 1000 block of E. Pine Street in Capitol Hill around 4:00 PM PST on February 16, 2023, then crashed the stolen car at Broadway and E. Cherry and fled on foot onto the Seattle University campus. SU Public Safety pushed a SeattleUAlert shelter-in-place that lasted approximately 90 minutes before being lifted.
2 messages in sequence · 2 verified verbatim
To check this alert, Claude (an AI) read it in full 25 separate times, independently. Each read decided whether the message answers each of the six questions and gave a short reason. A final reviewer then weighed all 25 and wrote the plain-English verdict you see when you open a row. The score (for example 22/25) is how many reads agreed; the 25 individual reads are tucked underneath if you want to check them.
We are sheltering in place until SPD gives an all clear. Card doors are locked.
Final assessment
A strong majority finds the source present, naming SPD as giving the all clear; two reads viewed SPD as a referenced third party, not the sender.
Who is sending the alert and who is responding. People act faster on a message from a clearly identifiable, credible sender, such as a named department, the police, or a branded alert system, than on an anonymous notice. A branded signature counts.
Final assessment
All 25 reads find no specific hazard named; the text only says they are sheltering without naming the threat.
What the threat actually is. A complete warning names the specific danger, such as a shooter, a fire, a tornado, or a gas leak, rather than a vague emergency, because people decide what to do based on what they are facing.
Final assessment
A majority finds no location named beyond locked Card doors; a sizable minority read Card doors as implying a building.
Where the threat is. Saying whether danger is in a specific building, a part of campus, or area-wide lets people judge their own proximity and choose a safe direction. Without a where, a warning is hard to act on precisely.
Final assessment
Unanimous that protective action is present; the message states they are sheltering in place with doors locked.
The protective action to take. A clear, specific instruction, such as shelter in place, evacuate, avoid the area, or run-hide-fight, drives faster and more correct protective behavior than describing the threat alone.
Final assessment
Nearly all reads find no time cue; two counted until SPD gives an all clear as a timeframe, which the majority did not credit.
When the message applies. A timestamp, the word now or immediately, or a phrase like until further notice tells the reader whether the danger is current and how quickly to act.
Final assessment
Absent, unanimous. States sheltering in place until an all-clear but names no hazard and conveys no harm or severity.
What the hazard could do to the people in its path. Beyond naming the threat, a complete warning conveys its potential consequences or severity, such as that a tornado can level buildings or that a leak could be explosive, so recipients grasp how much danger they are in. Research on warning message content finds that a concrete impact statement helps people personalize their risk and act sooner.
Systematic AI judgments with visible reasoning, not human-validated codings.
About this analysisCampus Alert Archive. "Seattle University: Armed carjacking suspects flee onto campus, prompting a 90-minute shelter-in-place." Incident of February 16, 2023. Added May 2026; last updated July 2026. https://campusalertarchive.com/case/seattle-university-carjacking-shelter-2023-02-16/
Alert text quoted on this page remains the work of the issuing institution; the archive is a secondary source.