Bank of Canada Governor Tiff Macklem said that Anthropic's new AI model Mythos creates serious cybersecurity threats to the financial system, urging swift regulatory action[2].
The warning matters because a breach could destabilize markets in Canada and abroad, undermining confidence in banks, payment networks, and investors worldwide[1].
Macklem said in New York after the International Monetary Fund meetings and during a conversation with Federal Reserve Chair Jerome Powell that the risks are not confined to domestic borders[3].
Macklem said that Mythos has already uncovered thousands of cybersecurity vulnerabilities in financial software, a pace and scale that outstrip traditional testing methods[5].
"I don't think anybody knows the full implications at this point," Macklem said, underscoring the uncertainty surrounding the model's capabilities[1].
"We need to find a way to protect the financial system, both within Canada and outside," he said, urging governments, and regulators to develop rapid‑response safeguards[1].
The Bank of Canada’s benchmark lending rate currently sits at 2.25%[6], a figure that could be affected if AI‑driven attacks force tighter credit conditions.
**What this means** The governor’s alert highlights a growing convergence of AI innovation and financial‑sector risk. If unchecked, advanced models like Mythos could become tools for cyber‑criminals to exploit system flaws faster than defenders can patch them, prompting regulators worldwide to consider new oversight frameworks for AI in finance.
“I don't think anybody knows the full implications at this point.”
Macklem’s warning signals that AI‑driven cyber threats are moving from theoretical to practical, pushing policymakers to craft rules that keep financial infrastructure secure while still fostering innovation.
