Canada is vulnerable to Anthropic's Mythos, says Conservative industry critic - The Logic

MP Raquel Dancho said she's glad the minister for AI appears to be taking the potential cybersecurity threat seriously, but said Canada still does not have the updated laws and frameworks to cope with the risks new AI models, including Mythos, could create. (The Logic)

Talking point: Anthropic claims its Claude Mythos model is capable of exploiting undiscovered vulnerabilities in software used for all kinds of critical security and infrastructure. Dancho said the government has yet to release its AI strategy, and the previous Liberal government's attempts to regulate AI were a "disaster" that never made it through Parliament. The government's updated cybersecurity law is still making its way through the Senate. AI Minister Evan Solomon is expected to release his long-awaited, updated AI strategy in the next few weeks, and said he feels confident in the security of Canada's institutions. OpenAI, meanwhile, announced that it has expanded access to its new advanced cybersecurity capabilities to thousands of verified individuals and hundreds of teams responsible for defending critical software.