TD Bank CEO Raymond Chun speaks at the lender's annual shareholders' meeting in Toronto on Thursday.Carlos Osorio/Reuters
Toronto-Dominion Bank TD-T chief executive officer Raymond Chun says continued collaboration between AI company Anthropic, government and businesses will be essential to avoid risks as artificial-intelligence technology rapidly evolves.
Banks, governments and regulators globally are assessing threats linked to the San Francisco-based Anthropic’s powerful Claude Mythos model. Mr. Chun said Mythos presents a new element of risk in AI adoption and the bank is working closely with the company and government to address cybersecurity concerns.
While TD does some work with Anthropic, the bank is not part of the group with access to a limited preview, but it will review the lessons that come out of the pilot, he said.
Opinion: Mythos sets the world on edge. What comes next may push us beyond
“I will commend Anthropic because they’re being very responsible in how they’re launching any new models into the public domain, and they’re starting with making sure that any of the major vendors and players are given the appropriate amount of time to do the vulnerability testing that’s required,” Mr. Chun said in an interview with The Globe and Mail.
“It’s important that everybody works together because the systems are all integrated.”
Mythos is skilled at reasoning and writing computer code, but it is also capable of finding and exploiting weak spots in software that could open companies to cyberattacks. Anthropic has restricted the pilot to test its Mythos model to select large companies, including Amazon and Microsoft, as well as JPMorgan Chase, the largest bank in the U.S.
After meeting with Anthropic officials on Tuesday, federal Artificial Intelligence Minister Evan Solomon said the company’s decision to not publicly release Mythos is the “responsible” approach.
On Friday, Canadian bank executives and regulators convened to discuss issues related to Mythos after U.S. regulators called a similar meeting with U.S. lenders. A Bank of Canada spokesperson said it was not an emergency meeting.
“We have invested significantly over the last number of years and are prepared that there will be different cyber issues that show up. Where Anthropic and Mythos are coming in is they’re bringing something different from a risk perspective,” Mr. Chun said.
“When it comes to cyber, there’s tremendous co-ordination and co-operation across industries, governments and regulators. It’s early days for Mythos, but we’re taking it very seriously.”
Andrew Willis: Canadian companies need access to Anthropic’s Mythos before hackers arrive
In September, just seven months after Mr. Chun took the helm as CEO, TD launched a strategic overhaul aimed at growing the bank’s businesses as it addresses gaps in its anti-money-laundering (AML) practices in the U.S.
Canada’s second-largest lender has been investing heavily in new AI tools to drive revenue and help employees do their jobs faster.
In the next three to four years, the banking sector will undergo significant transformation because of AI, Mr. Chun said. While AI has been used to lower costs and boost productivity, it is beginning to drive revenue.
For example, TD has cut the time it takes to assess mortgage applications from 15 hours to about three minutes. The process allows employees to deliver loans faster and spend more time having conversations with customers.
The bank is bulking up its front-line staff in Canada, adding 1,000 employees in branches, 1,200 wealth advisers, as well as hundreds of mortgage, investment and business specialists, according to Mr. Chun. TD plans to provide each employee with a personal AI agent that would compile a summary of a client’s portfolio, previous conversations and recommendations for products and services that better fit their needs.
“This is going to be a dynamic space,” Mr. Chun said. “The more we talk about it with our colleagues, the more they are seeing the benefits and the upsides.”
Opinion: Did anyone notice that AI companies have created HAL 9000?
Since U.S. regulators and law enforcement levied severe penalties on the bank in October, 2024, TD has been addressing strict remediation requirements. The bank has rolled out the first phase of a new platform to evaluate customer activity and enhanced customer screening procedures.
TD is continuing to conduct reviews of its previous AML practices and upgrade training for employees, among other measures. As part of this, TD is conducting more consistent assessments of its AML capabilities and systems and ensuring that they meet or exceed standards across the banking industry – a process that previously wasn’t being done sufficiently.
The bank has also hired new leaders and staff with experience in regulation, law enforcement and anti-money-laundering and has increased leadership training for executives across the company.
“We’ve made significant changes across other disciplines, compliance, legal team, and so on, making sure that you’ve got the best talent that is curious, that has the courage to escalate issues when you see them, and then hold people accountable,” Mr. Chun said.