CRA says AI isn't making tax decisions—so how is the agency using it?

As artificial intelligence becomes increasingly integrated into business and government operations, taxpayers should pay careful attention to its use by the Canada Revenue Agency (CRA).

While the CRA stated that it does not use AI to review or assess tax returns, the agency’s broader use of emerging technologies carries critical implications for taxpayers.

AI, automation, data analytics and business intelligence are particularly relevant to how the CRA processes information, assesses compliance risk and directs its review activity—so it’s imperative for taxpayers to understand the ongoing evolution of tax administration in Canada.

What the CRA says—and what it doesn’t say

The CRA says it currently uses AI in several of its operational and administrative functions, including: 

• Client services: The CRA’s chatbot, Charlie, was upgraded to a generative AI version in 2025. The chatbot provides general tax-related information to taxpayers, but does not process individual tax files or address personal tax situations.
  
  
• Administrative processes: Automation is used to help process high volumes of routine work, such as tax credits and rebates.
  
  
• Internal operations: AI tools assist with tasks such as document classification, drafting and summarizing internal documents, and supporting software development activities. 

The CRA emphasized that all AI results remain subject to human review by agency personnel and that generative AI tools are strictly prohibited for use on personal information.

Additionally, it said AI is limited to “rules-based algorithms” for risk-assessment purposes; the system follows preset if/then rules rather than learning or adapting on its own.

Implications for audit risk

While the CRA is relatively transparent about how AI supports its internal operations and taxpayer service, there is less information available about AI’s role in compliance activities.

For the 2025–2026 fiscal year, the CRA intends to: 

• Target high-risk groups for non-compliance using data analytics and business intelligence to identify aggressive tax planning.
• Enhance the use of technology—including machine learning and AI—to detect non-compliance and other suspicious activities.
• Use these tools to support early detection, prevention, verification measures and, where appropriate, criminal investigations.

The CRA’s focus on non-compliance is also reflected in recent measures included in Bill C-31 that would strengthen the agency’s audit and information-gathering powers.

For taxpayers, this could mean that data analytics and AI-supported processes may play a greater role in identifying higher-risk files, focusing compliance resources and shaping requests for information.

While the CRA has not yet publicly described the full extent of how these technologies are used in audit selection, taxpayers should anticipate that audits, reviews and information requests may be informed by automated tools that identify patterns, anomalies or other risk indicators—alongside human review.

What this means for taxpayers

As the CRA’s usage of data analytics and AI becomes more sophisticated, taxpayers should expect inconsistencies or unusual items to receive closer attention.

Potential areas of focus may include:

• Significant deviations from historical reporting patterns.
• Inconsistencies across multiple data sources.
• Digital asset activity, including cryptocurrency transactions.
• Complex and offshore structures.

There may be some ambiguities for taxpayers to navigate when it comes to specific risk indicators that contributed to the CRA’s decision to select a file; this adds further difficulty to understanding the basis of the agency’s review. As a proactive measure, taxpayers should maintain high-quality records and consistent reporting.

To ensure they are well-positioned to respond to CRA questions, taxpayers could consider reviewing whether their documentation clearly supports:

• The business purpose of significant transactions. 
• Material deductions, credits and incentive claims. 
• Significant changes from prior-year reporting. 
• The assumptions, estimates or judgements underlying key tax positions. 
• Reorganizations, restructurings or related-party transactions. 
• Differences across tax filings, financial statements, tax filings and other records.

Depending on a taxpayer’s specific situation, the appropriate advisors can help assess compliance risk and identify areas where additional support may be beneficial.