Canada is now on the exclusive list of countries and organizations with access to Anthropic’s Mythos AI model, which the company has claimed is too powerful to release to the wider public.
The news: US AI company Anthropic expanded the list of 50-odd organizations that have access to its cybersecurity model, Claude Mythos Preview, to roughly 200 participants on Tuesday. Federal AI and digital innovation minister Evan Solomon confirmed in a statement to BetaKit that the Canadian government now has access through the Canadian Centre for Cyber Security, a national cybersecurity agency under Communications Security Establishment Canada.
From the source: “This is an important step to help Canada’s cyber defenders better understand vulnerabilities, test systems responsibly, and strengthen protections for government services, critical infrastructure, data, and Canadian institutions,” Solomon’s statement said.
Following the thread: The Project Glasswing expansion comes a day after Anthropic confidentially filed to go public. Despite its insistence that its tactics are to prevent Mythos from getting into the hands of hackers, some have criticized Anthropic’s decision to limit access as a marketing ploy to convince customers that its tech is indispensable.
The company declined to share what specific Canadian organizations are on the list, but an Anthropic spokesperson said it’s limiting participants to organizations managing critical infrastructure like power, water, and healthcare, in addition to national security.
Final thought: Jon Ferguson, the vice-president of cyber and DNS at the Canadian Internet Registration Authority, told BetaKit last month that though Mythos may be more clever marketing than a technological revolution, it still showcases Canada’s vulnerability if it doesn’t have its own sovereign AI capacity.
Even with Project Glasswing, Canada is in a position where it’s reliant on the will of a US company to access an ultra-powerful AI model—just as it hopes to assert sovereign AI capacity through its upcoming AI strategy. A draft of the strategy, set to drop later this week, reportedly details plans to “build a multilateral alliance” to have sovereign autonomy in AI capabilities, per the CBC.
Feature image courtesy Anthropic via Flickr. Image license under CC BY 2.0.