Federal Advisory Body Recommends National Security Agencies Provide Transparency on AI Usage
A federal advisory body is urging Canada’s security agencies to publish detailed descriptions of their current and intended uses of artificial intelligence systems and software applications.
The National Security Transparency Advisory Group is also recommending that the government consider amending legislation to provide oversight of federal agencies’ use of AI, according to a new report.
Established in 2019 to enhance accountability and public awareness of national security policies, programs, and activities, the group is seen as a significant tool for implementing a federal commitment to transparency.
While security agencies emphasize the importance of openness, they acknowledge limitations on what they can publicly disclose due to the sensitive nature of their work.
The report highlights the increasing use of AI by security agencies for various tasks, including document translation and malware detection. It anticipates a growing reliance on AI for tasks like text and image analysis, pattern recognition, and behavior interpretation.
As the use of AI grows in the national security community, there is a pressing need for increased public understanding of the goals and activities of national border, police, and intelligence services, the report emphasizes.
The report calls for the establishment of mechanisms to enhance government openness and facilitate external oversight and review.
One challenge in explaining AI to the public is the opacity of algorithms and machine learning models, leading to a lack of understanding even among security agencies about their own AI applications.
Ottawa has provided guidance on federal AI use, requiring an algorithmic impact assessment before deploying systems that support or replace human decision-making.
The proposed Artificial Intelligence and Data Act aims to ensure responsible design, development, and implementation of AI systems, but the act does not cover government institutions like security agencies, prompting the advisory group to suggest its extension.
While emphasizing the essential role of humans in the AI process, the report highlights the Canadian Security Establishment’s efforts in using AI, such as training new models and developing translation tools.
The report also mentions the Cyber Centre’s use of AI tools to detect phishing campaigns and suspicious activities on federal networks.
The transparency advisory group acknowledges the challenges security agencies face in disclosing details of their AI use due to their mandate to safeguard national security.
Despite these limitations, security agencies are working on ethical approaches to govern, manage, and monitor AI usage, drawing on best practices and dialogue for guidance.
The report notes that while transparency is crucial for innovation and public trust, there are limitations on what can be publicly shared by agencies like the Canadian Security Intelligence Service to avoid compromising operations.
In response to concerns over facial recognition technology, the RCMP introduced a Technology Onboarding Program to assess compliance with privacy legislation. The transparency group urges the RCMP to provide more information about this program to improve transparency.
The RCMP plans to publish a transparency blueprint outlining the onboarding program’s key principles and assessed tools, as well as developing a national policy on AI use with transparency measures.
The transparency advisory group also calls for more public reporting on the government’s transparency initiatives and recommends a formal review with public reporting of progress, impacts, and future activities.
Public Safety Canada has shared the report’s recommendations with relevant stakeholders but has not confirmed agreement or provided a timeline for implementation.