AI-Assisted Facial Recognition for Arson Investigations

What it is for

Enforcement

Used by the Bureau of Fire Investigations to assist in identifying individuals suspected of arson and other criminal activities that fall within the Fire Department's jurisdiction.

NYC Algorithmic Tools Compliance Report

Run by: Fire Department (FDNY)
Where: No fixed location
Kept: Not stated by the NYC Office of Technology and Innovation.
Shared with: Accountable organization
Your copy: You cannot see the data it holds about you. What you can do

What it is for

Enforcement

This system is used by the New York City Fire Department's Bureau of Fire Investigations to help identify individuals in arson and criminal fire cases. It compares uploaded images or video frames against a database of lawfully held images to generate a list of possible matches. Any matches are treated only as investigative leads — Fire Department personnel must conduct additional investigation before any identification is acted upon, and a match alone is never treated as a confirmed identification.

What it collects and what happens to it

Data taken in

Biometric

Identifiable data

Images of individuals of interest uploaded by investigators, as well as images sourced from social media or the internet. These include facial images used as query inputs to the recognition system.

NYC Algorithmic Tools Compliance Report

Sensitive personal information

Identifiable data

Training data consists of publicly available open-source social media images of individuals. The reference database against which queries are compared is composed of lawfully possessed images.

NYC Algorithmic Tools Compliance Report

Processing

Biometric Recognition

The core processing technique is facial recognition: the system extracts facial feature embeddings from query images and compares them to a reference set to rank candidates by similarity. This is a biometric comparison against named or identifiable individuals in the reference database.

NYC Algorithmic Tools Compliance Report

Computer Vision

The system also performs general computer vision to process uploaded video frames and images — extracting faces, normalizing them, and preparing them for biometric comparison.

NYC Algorithmic Tools Compliance Report

What it does

Deciding (Analytical AI)

Human decides

The system classifies and ranks uploaded images against a database of known images, generating a pool of possible matches scored by similarity. All match results are advisory — Bureau of Fire Investigations personnel review results and conduct additional investigation before treating any output as an identification.

NYC Algorithmic Tools Compliance Report

Sensing (Perceptive AI)

Human decides

The system ingests raw images and video frames, extracting facial features and converting them into structured representations that are then compared against the reference database.

NYC Algorithmic Tools Compliance Report

Outputs

A recommendation or prediction

Identifiable data

The system outputs a ranked pool of possible image matches — candidate individuals whose images are similar to the query. These results serve only as investigative leads and are not treated as positive identifications. Bureau personnel must conduct independent follow-up investigation before acting on any match.

NYC Algorithmic Tools Compliance Report

Run by

Fire Department (FDNY)

The Fire Department's Bureau of Fire Investigations deploys and operates this facial recognition tool to support investigations into arson and other criminal activities within the agency's jurisdiction. Bureau personnel review all match results before any investigative action is taken.

NYC Algorithmic Tools Compliance Report

Built by

Undisclosed Off-the-Shelf Facial Recognition Vendor

The vendor provides an off-the-shelf facial recognition product and ongoing technical support. The vendor's identity is not publicly disclosed; confidentiality agreements are in place between the vendor and the Fire Department.

NYC Algorithmic Tools Compliance Report

Kept for

Not stated by the NYC Office of Technology and Innovation.

Shared with

Available to the accountable organization

Match results and query outputs are available to Bureau of Fire Investigations personnel within the FDNY for use as investigative leads in arson and criminal fire cases.

NYC Algorithmic Tools Compliance Report

Not available to me

Individuals do not have access to match results or information about whether they have been searched in this system. The vendor is also not publicly disclosed.

NYC Algorithmic Tools Compliance Report

Stored

Not stated by the NYC Office of Technology and Innovation.

How to read the colours

Can it identify you?

Anonymized data: Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
Pseudonymous data: Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
Identifiable data: The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.

Who completes the loop?

Human decides: This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
Human executes: This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
Autonomous: This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.

Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.

What you can do

Ask about this system

Questions go to the NYC Office of Technology and Innovation, not the vendor.

Your rights

Right to Be Informed of AI UseThe existence of this facial recognition tool is disclosed through the NYC Algorithmic Tools Compliance Report, a public register. Individuals who may have been subjects of a facial recognition search as part of a fire investigation are not individually notified that the tool was used; however, the tool's existence and general operation are publicly disclosed through this registry.
Right to a Human ReviewBureau of Fire Investigations personnel conduct an additional investigation for any match the system generates. No match is treated as a positive identification without human review. Individuals who become subjects of an investigation resulting from a match retain the right to contest findings through standard law-enforcement and legal processes.

Risks and safeguards

Reputational harmA false positive match could cause an innocent person to become a suspect in an arson investigation, leading to reputational damage, wrongful questioning, or arrest.Safeguard: The register states that any match serves only as an investigative lead, and Bureau of Fire Investigations personnel must conduct additional independent investigation. The match alone is explicitly not treated as a positive identification.
Civil liberties harmUse of facial recognition in criminal investigations raises risks of civil liberties infringement, including potential chilling effects on lawful assembly and movement, and disproportionate impact on communities already over-surveilled. Training on social media data may reflect demographic biases.Safeguard: The tool is limited in scope to arson and criminal fire investigations within FDNY jurisdiction. Human review is required before any match is acted upon. The vendor identity and technical details are not publicly disclosed, limiting independent audit.