From the Emergency Services edition of November 4, 2026
Seattle Times: 2019 city privacy memo flagged equity risks of Fire Department AI on 911 medical calls, but live AI prompts launched in 2023 without public review
Seattle Fire Department; City of Seattle Information Technology Department; Corti (vendor) · 911 medical call triage and EMS dispatch · Seattle, Washington
- Publisher
- The Seattle Times
- Original publication
- 2026-11-04
- Source retrieved
- 2026-11-05
What happened
Public records obtained by The Seattle Times show Seattle's then chief privacy officer wrote a 2019 privacy risk analysis warning that Corti AI on the Fire Department's 911 call-taking platform could raise equity concerns tied to language, accents and disabilities, and recommended a race and social justice assessment and public disclosure. The project was framed as a quality-assurance pilot on completed calls and skipped Seattle's normal purchasing and full technology review. Since 2023 AI has listened to all 911 medical calls and prompts dispatchers on routing some callers to a nurse line. The department says human dispatchers stay in control; it has asked IT for retroactive privacy reviews, and UW Medicine is evaluating the AI and nurse line. Experts and ACLU of Washington urge suspending live prompts pending a public audit; the mayor's office promised an update on city AI use in early 2027.
Why it matters
A concrete US PSAP governance failure: a vendor AI moved from pilot QA to live 911 triage prompts without the review a city policy officer recommended. Relevant to any fire-based dispatch or 911 authority adopting AI call analysis, and to state 911 offices setting AI policy. Interpretation: pilot-to-production drift is the control gap, not the model alone.
Evidence and measured results
Investigative newspaper report based on public records and interviews; the 2019 memo is quoted. The department has shared no equity or accent performance data; the cited Danish stroke-detection study does not cover nurse-line triage. No outcome data on AI-routed callers reported.
Limitations and uncertainty
Single newspaper investigation; department disputes framing; no measured disparity data either way; UW Medicine evaluation scope and timing not disclosed.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-05; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Agencies running AI on live 911 calls lack documented evidence it performs equitably and was properly reviewed.
- Stakeholders
- Fire chief
- EMS medical director
- City chief privacy officer
- 911 director
- City council public safety committee
- Discovery questions
- Which AI functions run on live calls today?
- Were pilots re-reviewed before going live?
- Are AI prompts logged alongside dispatcher decisions?
- Has performance been tested across accents and disabilities?
- Value hypothesis
- Interpretation: an independent assessment could restore public trust and reduce legal and equity risk of existing AI use.
- Potential engagement
- A retroactive AI governance and equity assessment of 911 call-analysis tools.
- Unsupported claims
- That the AI caused patient harm
- Measured bias in Corti models
- Improved triage accuracy in Seattle
Pre-sales engineering
Role takeaway
- Fit
- Fits PSAPs and fire-based dispatch centers using vendor AI on live call audio.
- Architecture and integration
- Vendor speech AI embedded in call-taking platform analyzing live audio and prompting dispatchers; downstream nurse-line routing.
- Prerequisites
- Prompt and decision logs
- Access to vendor performance data
- Data agreement covering audio
- Constraints
- Vendor-controlled models
- No subgroup performance data
- Sensitive caller audio
- Security
- Confirm audio storage location, retention and vendor access; log every AI prompt and override.
- Proof of value
- Compare AI prompts with dispatcher outcomes
- Test speech accuracy across accent and disability samples
- Audit nurse-line routed outcomes
Delivery
Role takeaway
- Work
- Inventory AI use, run retroactive privacy and equity reviews, publish findings and set a re-approval gate.
- Dependencies
- Vendor cooperation
- City IT privacy office
- Medical director
- Ownership
- Fire chief with EMS medical director and city privacy office.
- Skills and adoption
- Dispatchers trained on when to disregard AI prompts and how to flag errors.
- Governance checkpoints
- Privacy impact review
- Race and social justice assessment
- Public disclosure and council briefing
- Acceptance criteria
- Proposed: published AI use inventory
- Proposed: subgroup performance results disclosed
- Proposed: re-approval required before scope changes
- Risks
- Missing historic logs
- Vendor non-cooperation
- Public trust erosion
- Automation bias among dispatchers
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
AI speech analysis runs on live 911 audio inside the vendor call-taking platform and surfaces prompts to dispatchers; performance depends on speech recognition across accents and impairments.
Governance
Who approves, reviews and stays accountable for outcomes?
Pilots that touch live calls need re-review when scope changes; public disclosure, equity assessment and audit trails of AI prompts versus dispatcher decisions are the requested controls.
Security and privacy
What data, permissions and controls need testing?
Live 911 audio shared with a foreign vendor raised data ownership, sharing and breach concerns, partly addressed in a 2019 data agreement; retroactive privacy reviews are now underway.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Callers with disabilities, accents, stress or limited English may be understood less well; dispatchers may over-defer to AI prompts.
Procurement
What should contracts, pricing and exit terms secure?
The deployment bypassed normal purchasing as a pilot. Interpretation: contracts should require subgroup performance data and re-approval before moving to live use.
Operating model
Which teams own the service once it runs?
AI prompts feed diversion of lower-acuity callers to a nurse line exempt from response-time standards, making triage errors consequential.
Publication history
- 2026-11-04Emergency Services · Issue 593 resources
Stable resource ID: seattle-times-sfd-corti-ai-911-equity-memo-ignored-20261104