The provincial Ministry of Technology and Innovation (MITI) developed a system powered by Anthropic’s technology to test its applications against security and privacy standards, and target vulnerable ones for fixing or replacement. The tools used commercially available AI models Claude Opus and Sonnet. (The Logic)

Talking point: Every government across the country has a bunch of legacy applications and software systems they’ve needed to modernize for a while. Alberta has centralized responsibility for the technology of provincial departments under MITI, led by Technology Minister Nate Glubish, a former venture capitalist who’s bullish on the potential of AI in the public service. The province claims AI tools can help cut the cost of technology modernization to just five per cent of the $2 billion it would otherwise take. Anthropic touted the results announced Monday as an example of how governments can employ widely used versions of Claude to protect their systems; it comes as policymakers have expressed concern about the cybersecurity threat the San Francisco-based firm claims its Mythos model could pose. The federal and Quebec governments, meanwhile, have turned to Canadian startup Cohere for their AI tools.