AI-Assisted Risk Scoring for Inbound Air Cargo Security

Safety & Security · Risk Assessment & Triage · Enforcement

What it collects that can identify you

Sensitive personal information

Identifiable data

Shipper and consignee names and addresses submitted on air waybills by air carriers via web API. These are directly identifying personal data protected at the Protected A classification level under Canadian law.

PACT AIA — Section 3, Q43–Q54

Also collects operational data, which is anonymized data.

Run by: Transport Canada (TC)
Where: No fixed location
Kept: Retained As per applicable Canadian laws governing PLACI data retention under the Canadian Aviation Security Regulations, 2012
Shared with: Accountable organization
Your copy: You cannot see the data it holds about you. What you can do

What it is for

Safety & Security

Risk Assessment & Triage

Enforcement

This system automatically scores every inbound air cargo shipment to Canada for security risk before it is loaded onto an aircraft. It uses machine learning and natural language processing to analyse air waybill data — including shipper names, addresses, and commodity descriptions — and flags high-risk shipments for review by Transport Canada analysts. Low-risk shipments are cleared automatically; higher-risk ones are reviewed by a human analyst who decides whether to request more information, require additional screening, or issue a do-not-load notice.

What it collects and what happens to it

Data taken in

Sensitive personal information

Identifiable data

Shipper and consignee names and addresses submitted on air waybills by air carriers via web API. These are directly identifying personal data protected at the Protected A classification level under Canadian law.

PACT AIA — Section 3, Q43–Q54

Operational data

Anonymized data

Commodity description, weight, piece count, and air waybill number — the non-personal operational fields from air waybills submitted by air carriers. These follow a mandatory industry-standard format transmitted through a web API.

PACT AIA — Section 3, Q54

Processing

Classification & Prediction

Supervised machine learning classifies and groups free-text data fields in air cargo information and applies an automated risk algorithm to calculate a risk score for each inbound shipment. The algorithm was developed from Transport Canada's own training data.

PACT AI Register — Description

Language Models

Natural language processing interprets free-text commodity description fields on air waybills, which are rife with typos and industry jargon. NLP normalizes these descriptions and conducts identity resolution on shipper and consignee names.

PACT AI Register — Description

What it does

Deciding (Analytical AI)

Human decides

The system calculates a risk score for each shipment and classifies it as low-risk (auto-assessed) or above-threshold (referred to a human analyst). For above-threshold shipments, the system displays the score and relevant trade-pattern information; a human PACT analyst makes the final risk-mitigation decision.

PACT AIA — Section 3, Q22–Q23

Understanding (Semantic AI)

Human executes

Natural language processing and fuzzy matching interpret free-text fields in air waybills — which are rife with typos and industry jargon — grouping commodity descriptions into meaningful categories (e.g. jeans/denim/pants = clothes) and performing identity resolution on shipper and consignee names.

PACT AIA — Section 3, Q25

Outputs

A recommendation or prediction

Identifiable data

An inbound cargo risk score displayed to PACT analysts for each shipment, along with relevant trade-pattern information. This score delineates shipments requiring further human review from those that do not. The score is tied to individual shipments associated with named shippers.

PACT AIA — Section 3, Q26

A decision about you

Anonymized data

Electronic status messages sent to air carriers: error codes for incomplete data, assessment-complete messages for low-risk shipments, and (following human review) RFI, RFS, or Do Not Load notices. Low-risk auto-assessments are system-generated without personal identifiers beyond the waybill number.

PACT AIA — Section 3, Q26

Run by

Transport Canada (TC)

Transport Canada is the federal department accountable for deploying and operating the PACT system. The system is restricted to specific users on the Transport Canada intranet, and Transport Canada analysts make all final risk-mitigation decisions.

PACT AI Register Entry

Built by

The Employment Solution, TPG Technology Consulting, ADRM Technology Consulting Group, Microsoft

A consortium of vendors built the PACT AI system on behalf of Transport Canada. Microsoft provides underlying cloud and AI platform capabilities. The other vendors contributed system integration, technology consulting, and domain-specific design.

PACT AI Register Entry — Vendor field

Kept for

Retained As per applicable Canadian laws governing PLACI data retention under the Canadian Aviation Security Regulations, 2012

Personal information collected is stored and disposed of in keeping with all applicable laws in Canada. A Privacy Impact Assessment is underway to review collection, use, disclosure, and retention of PLACI data. The system will not de-identify personal information used or created.
Duration: As per applicable Canadian laws governing PLACI data retention under the Canadian Aviation Security Regulations, 2012

PACT AIA — Section 3.2, Mitigation Q34–Q35 and Q39

Shared with

Available to the accountable organization

Risk scores, shipment data, and audit trail records are available to authorized Transport Canada PACT analysts and managers through the PACT system on the Transport Canada intranet. Access is controlled by role-based permissions.

PACT AIA — Section 3, Q6 and Q27

Not available to me

Individual shippers and consignees do not have access to the risk scores assigned to their shipments or the internal assessment data held by the PACT system. The system operates on the Transport Canada intranet and is restricted to authorized government employees.

PACT AIA — Section 3, Q6

Stored

Stored on 3rd Party Cloud

Microsoft is listed as a vendor, indicating cloud infrastructure is used. Security and privacy have been designed into the system from the concept stage, and sharing agreements with appropriate safeguards are in place.
Duration: As per applicable Canadian laws governing PLACI data retention

PACT AI Register — Vendor field (Microsoft) · PACT AIA — Section 3.2, Mitigation Q36 and Q38

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 Helpful Places, not the vendor.

Your rights

Right to Be Informed of AI UseA plain language notice is required to be posted through all service delivery channels in use (internet, in person, mail, or telephone) informing affected parties that an automated decision system is in operation. The AIA is publicly available on the Government of Canada Open Data portal.
Right to an Explanation of a DecisionWhere a decision results in denial of a service or a regulatory action (e.g. a Do Not Load notice), shippers are entitled to a meaningful plain-language explanation covering: the role of the system in the decision, the data and its source, the criteria applied, the system output and how to interpret it, and the principal factors that led to the decision. Contact information for the PACT Program is provided through which shippers can request an explanation.
Right to ContestA recourse process is established for clients who wish to challenge a PACT decision. Shippers may invoke their rights under their contract of carriage or under international treaties governing liability for the shipment of goods on aircraft. Contact information for the PACT Program is provided to shippers when a Do Not Load decision prevents their cargo from being loaded.
Right to a Human ReviewAll shipments above the low-risk threshold are reviewed by a human PACT analyst before any enforcement action is taken. The system enables human override of system decisions, and a log is maintained of all instances when overrides are performed. Low-risk auto-assessments do not involve human review, but no enforcement action is triggered by auto-assessment alone.
Right to Non-discriminationA Gender-based Analysis Plus was conducted to identify and address potential discriminatory impacts on gender and other identity factors. The use of advanced analytics is expected to significantly reduce the potential for unintentional bias compared to manual assessment. Documented processes are in place to test datasets against biases, though these are not publicly available.

Risks and safeguards

Civil liberties harmThe system could produce discriminatory targeting based on shipper names, nationalities, or trade routes, affecting individuals' freedom from arbitrary interference.Safeguard: A Gender-based Analysis Plus was conducted; documented processes test datasets against biases; the risk algorithm is transparent and not a trade secret; all key decision points are linked to relevant legislation; and human analysts make all final enforcement decisions, with a recourse process available to shippers who wish to challenge outcomes.
Financial & business harmA shipper whose shipment is re-screened or prevented from loading could incur additional costs and supply-chain delays.Safeguard: The system is designed to minimize operational impacts; air carriers are committed to expedient risk assessment; impacts are assessed as typically brief and limited to a single shipment; and shippers have recourse through their contract of carriage and international liability treaties.