The Globe and Mail reports in its Thursday, Oct. 8, edition that Canada's banking regulator plans to implement tighter artificial intelligence guidelines amid rising risks from AI agents. The Globe's Stefanie Marotta writes that as banks rapidly adopt AI to reduce costs and enhance returns, Office of the Superintendent of Financial Institutions superintendent Peter Routledge plans to develop a "safety code" to improve cybersecurity and mitigate these risks, as he said at a Global Risk Institute conference. The regulator is still refining the details, but Mr. Routledge said the safety code may include broad standards to ensure basic safety for the financial system while allowing room for innovation. Rather than implement a regulatory framework prior to now, OSFI has guided banks to "do no harm," Mr. Routledge said. The introduction of Mythos, which Anthropic has restricted the use of because it is capable of finding and exploiting weak spots in software that could open companies to cyberattacks, was the catalyst that spurred OSFI to consider firmer guardrails on the banking industry's use of the technology. Mr. Routledge said Mythos's software exploiting abilities "constituted a major increase in cyber risk."