tech

Bell and Cohere train a Canadian AI model to investigate cyberattacks

The system draws on nearly a decade of Bell Cyber data to help analysts move faster during incidents.

The Canadian

Technology

September 23, 2026

5 min read

Cybersecurity analysts monitoring blue data screens

Bell Canada and Toronto-based Cohere have developed a customized language model intended to speed up cybersecurity investigations. The system was trained over seven months using nearly a decade of operational data from Bell Cyber.

The model is designed to summarize large volumes of incident information and help analysts identify patterns more quickly. Bell Cyber serves major organizations, including CIBC and the federal and Ontario governments.

Bell Cyber chief executive John Menezes said defenders increasingly need AI because attackers are also using automated tools. The company says the goal is not to remove analysts, but to shift their time from repetitive synthesis toward higher-level investigation and judgment.

The project is a significant test of Canadian enterprise AI in a sensitive environment. Its credibility will depend on accuracy, safeguards around client data, and clear human accountability when automated analysis shapes a security response.

Related coverage