Skip to content
This is NOT an official site of the Government of Canada. Click here for the official AI registry.

Helpful Places

Government of Canada – DTPR AI Register

Back to your search

AI-Assisted Modernization of Legacy Mainframe Code

Research & Development · Planning & Decision-making

What it collects

Operational data
Anonymized data
  • Legacy COBOL mainframe source code and application artifacts from the CRA's existing mainframe systems. This is software code, not personal data about individuals.
Run by
Canada Revenue Agency (CRA)
Where
No fixed location
Kept
Not stated by the Helpful Places.
Shared with
Accountable organization, Vendor

What it is for

Research & DevelopmentPlanning & Decision-making

This system uses AI to analyze the Canada Revenue Agency's legacy COBOL mainframe code, automatically identifying key components, generating documentation, and extracting business logic. It is used internally by CRA employees to help plan and carry out the modernization of aging government software. The system is currently a proof of concept and does not affect citizen-facing services or decisions.

What it collects and what happens to it

Data taken in

Operational data
Anonymized data
  • Legacy COBOL mainframe source code and application artifacts from the CRA's existing mainframe systems. This is software code, not personal data about individuals.

Processing

Classification & Prediction
  • Uses supervised and unsupervised machine learning to automatically identify and classify key components in legacy COBOL code, extract business logic, and predict structure and relationships within the codebase.
Language Models
  • Language model capabilities within AWS Transform – Mainframe are used to generate human-readable documentation and extract natural-language descriptions of business logic from COBOL code.

What it does

Deciding (Analytical AI)
Human decides
  • The AI analyzes legacy COBOL code to classify and label key components, predict business rules, and rank areas of complexity. Human software architects and developers review the outputs and decide on modernization strategy.
Creating (Generative AI)
Human decides
  • The AI generates documentation from the analyzed COBOL code, producing written descriptions of program logic and component relationships that did not previously exist as structured documents. CRA staff review all generated documentation before use.

Outputs

Operational data
Anonymized data
  • Structured maps and inventories of COBOL application components, identified business rules, and dependency charts describing the existing mainframe architecture. These outputs are administrative artefacts for internal use only.
Generated content
Anonymized data
  • AI-generated documentation describing the logic, structure, and purpose of legacy COBOL programs. This content is reviewed by CRA technical staff before being used in modernization planning.

Run by

Canada Revenue Agency (CRA)
  • The Canada Revenue Agency (CRA) is the federal government department responsible for tax administration and benefit delivery. CRA is deploying this proof-of-concept AI system to support internal software modernization efforts.

Canada AI Register — 2526-CRA-ARC-009

Built by

Amazon Web Services (AWS)
  • Amazon Web Services provides the AWS Transform – Mainframe AI service used in this proof of concept. AWS supplies the underlying AI models and cloud infrastructure that analyze the COBOL codebase.

Canada AI Register — 2526-CRA-ARC-009

Kept for

Not stated by the Helpful Places.

Shared with

Available to the accountable organization
  • Outputs (code maps, documentation, extracted business logic) are available to CRA employees involved in the modernization project. This is an internal government tool with no public-facing output.
Available to vendor
  • CRA mainframe code is processed through the AWS Transform – Mainframe service. As a cloud service, source code and derived outputs may be accessible to AWS for service delivery purposes. The extent of vendor data access is not fully specified in the register entry.

Stored

Stored on 3rd Party Cloud
  • The system operates on AWS cloud infrastructure. CRA mainframe code and AI outputs are processed and potentially stored on Amazon Web Services infrastructure. Specific data residency and retention terms are not detailed in the register entry.
  • Duration: Not specified
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 Algorithmic TransparencyThis system processes internal government software code and is used only by CRA employees. Citizens are not directly subject to its outputs. Information about this system is publicly disclosed via the Government of Canada AI Register. The underlying AWS Transform – Mainframe service documentation is publicly available from Amazon Web Services.

Risks and safeguards

  • Reputational harmThe AI may incorrectly classify or document business logic from legacy code, potentially leading to misunderstandings of critical tax administration rules if outputs are acted on without review.Safeguard: The system is a proof of concept with outputs reviewed by CRA technical staff before informing any modernization decisions. No automated decisions about citizens are produced.
  • Civil liberties harmCRA mainframe code may encode sensitive tax administration business logic; if cloud-processed, there is a potential risk of unauthorized exposure of government system internals to a third-party vendor.Safeguard: The proof-of-concept scope is limited; the register entry does not detail what data-protection agreements govern the AWS processing relationship. Public disclosure through the AI Register supports accountability.