From the Emergency Services edition of September 14, 2026
Seattle Fire Corti EMS documentation pilot emphasizes HITL review and patient notice
City of Seattle / Seattle Fire Department · Municipal fire/EMS / city privacy governance · United States (Seattle, Washington)
- Publisher
- City of Seattle Privacy and AI Impact Assessment
- Original publication
- 2026-08-31
- Source retrieved
- 2026-09-15
- Event date
- 2026-08-01
What happened
Seattle's Aug 31, 2026 Privacy/AI Impact Assessment describes a multi-month Corti Assistant pilot in which transcription and an in-app LLM surface medically relevant fields for EMT/paramedic review before manual paste into ESO, with patient notice, opt-out, no ESO integration, stated no vendor training on Seattle data, and multiple human validation steps—documentation support, not autonomous triage.
Why it matters
Why now: newly relevant after September 2026 public scrutiny (already archived) of Seattle Fire AI on 911 medical calls; this PIA is a distinct, previously unarchived governance artifact for a field-documentation pilot that makes notice and HITL controls explicit, informing SLED delivery patterns for EMS AI.
Evidence and measured results
PIA sections detail scope (handwritten/typed notes vs transcription+LLM surfacing), validation sequence, consent/notice language, HIPAA/RCW framing, cloud processing claims, retention alignment, and explicit statement that the system produces recommendations and cannot act without human intervention. Anticipated benefits are not measured in the document.
Limitations and uncertainty
Self-assessment not independent audit; benefits unverified; does not evaluate Corti live triage prompting; pilot parameters may change.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-09-15; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- EMS clinicians lose time and accuracy when manually charting vitals and history under time pressure, but agencies face privacy and trust barriers to AI scribes.
- Stakeholders
- Fire/EMS medical director
- Field EMTs/paramedics
- City privacy/IT security
- Receiving hospital/ePCR administrators
- Discovery questions
- Will patients be notified and able to decline transcription?
- Is writeback to the ePCR in or out of scope for the pilot?
- Who attests the final chart if AI drafted fields?
- What retention and non-training terms are in the vendor contract?
- Value hypothesis
- A notice-enabled, HITL documentation assistant may improve chart completeness if correction burden stays low—without claiming Seattle has proven outcome gains.
- Potential engagement
- Privacy/AIA-aligned pilot design for one battalion with weekly fidelity audits and an explicit no-triage-scope clause.
- Unsupported claims
- Measured reduction in charting time at the prospect
- Clinical outcome improvement from transcription AI
- That this PIA authorizes or evaluates live 911 triage AI
Pre-sales engineering
Role takeaway
- Fit
- Field documentation transcription and LLM field-surfacing with human gatekeeping; not CAD triage or nurse-line diversion.
- Architecture and integration
- Managed mobile app; vendor cloud STT/LLM; manual copy into existing ePCR (ESO-class); no automated integration in the described Seattle pilot.
- Prerequisites
- City-managed devices only
- BAA and non-training contractual controls
- Patient notice script and opt-out logging
- QA sampling process for transcript errors
- Constraints
- Implied consent contexts for altered patients
- Cloud processing of PHI
- Pilot-only scope and retention rules
- Security
- Enforce role-based access, encryption, deletion after retention, and vendor employee access only with agency permission; verify controls beyond vendor assertions.
- Proof of value
- Clinician correction rate per chart
- Time from patient contact to completed ePCR fields
- Count of opt-outs and any PHI incidents
Delivery
Role takeaway
- Work
- Configure app on unit devices, train notice and review workflow, run multi-month pilot sampling, and decide expand/stop with privacy and medical directors.
- Dependencies
- Vendor provisioning
- City privacy review already exemplified by PIA
- ePCR process owners
- Field training time
- Ownership
- SFD/EMS clinical leadership owns chart quality; city privacy owns PIA compliance; vendor owns platform controls under contract.
- Skills and adoption
- Train crews to treat AI output as draft text; practice declining use when patients object; escalate chart conflicts to lead clinician.
- Governance checkpoints
- PIA publication/update
- Mid-pilot accuracy review
- Scope check barring triage features
- Retention/deletion verification
- Acceptance criteria
- Documented HITL attestation on sampled charts
- Working opt-out logging
- No unauthorized ePCR auto-write
- Risks
- Hallucinated meds/vitals entering charts
- Scope creep into call triage
- Public distrust if notice is skipped
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?
Keep documentation AI on managed devices with manual EHR paste (no silent integration) until accuracy and retention are proven; separate this stack from live 911 audio-prompting systems architecturally and contractually.
Governance
Who approves, reviews and stays accountable for outcomes?
Require published PIAs, patient notice/opt-out where feasible, clinician attestation before chart finalization, and clear distinction between documentation assistance and triage/diversion AI.
Security and privacy
What data, permissions and controls need testing?
PHI in vendor cloud demands BAAs, access logging, deletion assurances, and limits on employee access; verify no-training-on-customer-data contractually and technically.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
May reduce after-call documentation load if accurate; requires training so crews do not defer clinical judgment to surfaced fields; opt-out must work for patients who refuse recording-like tools.
Procurement
What should contracts, pricing and exit terms secure?
Use the PIA as a template checklist: BAA, non-training, managed devices, no silent EHR integration, and published privacy assessment before go-live.
Operating model
Which teams own the service once it runs?
Separate documentation AI governance from any live 911 prompting program; keep clinician attestation as the chart source of truth.
Publication history
- 2026-09-14Emergency Services · Issue 093 resources
Stable resource ID: seattle-corti-ems-documentation-pia-2026