It Takes Seconds: Canadian Cyber Leaders Say Your Staff’s AI Habits Are the Real Data Risk

The most dangerous cyberattack in your company this month probably did not come from a hacker in another country. According to Canadian technology and advisory leaders, it likely came from an employee trying to save a few minutes, pasting a client email or a spreadsheet of financials into a free AI chatbot. That single copy-paste can put customer data somewhere you will never get it back.

A new set of warnings from executives at Kyndryl Canada, Xero, BDO Canada and Wagepoint paints a blunt picture. AI is not just creating new kinds of attacks. It is making familiar ones faster, and it is quietly widening the door through everyday office habits.

The copy-paste habit that keeps security teams awake

Suzy Clarke, who leads security at the accounting software company Xero, put the internal risk in the starkest terms. She told SecurityBrief's Karen Joy Bacudo that many conversations about AI and cybersecurity focus on scammers, while some of the biggest risks come from seemingly harmless habits inside the business.

"It takes seconds to paste a client email or a set of financials into a free AI tool to save time, and depending on your settings, that data may be stored or used to train the model. Once it's out there, it's very hard to get back," Clarke said.

That is the crux of it. No one broke in. No firewall failed. A worker did something efficient, and the company lost control of sensitive material. Clarke's advice is deliberately unglamorous: agree on which AI tools staff may use, check the privacy settings on those tools, and write clear AI usage policies that keep customer and financial information out of public chatbots completely.

Attackers are not inventing new tricks. They are getting faster

The external side of the picture is just as sobering. Denis Villeneuve, who leads cyber resilience at Kyndryl Canada, said the biggest shift he sees is speed rather than novelty.

"AI isn't necessarily creating entirely new risks, but it can accelerate familiar ones and give attackers the ability to find and exploit vulnerabilities faster," Villeneuve said. "That changes the equation for cyber leaders."

Those playbooks assumed a certain timeline. Now AI can shorten the gap between discovering a weakness and exploiting it, which means companies have less time to respond than they planned for.

Villeneuve's prescription is practical: strong identity controls, visibility across the whole environment, and the ability to spot problems, fix them and recover quickly. His message was that staying secure requires ongoing attention and fundamentals that move at the same pace as the threat landscape.

Cybersecurity is now a boardroom subject

Perhaps the most telling shift is where these warnings are being aimed. Rocco Galletto, the national and global cybersecurity leader at BDO Canada, said cyber risk now sits firmly at the board level, and that is where it belongs.

"Cybersecurity is a business risk, and governing that risk is a boardroom responsibility," Galletto said. He added that directors do not need deep technical knowledge, but they do need evidence that resilience measures are in place and aligned with how the business actually operates.

That is a meaningful change in tone. For years, boards treated security as an IT line item, something the technical team handled quietly. Galletto's argument is that boards should understand what the company is exposed to, set a clear tolerance for risk, and ask hard questions about whether the business could respond if something went wrong: the board's job is to ask the right questions, demand clear evidence of resilience, and keep cybersecurity decisions aligned with business strategy.

The comments reflect a wider expectation, the reporting noted, that cyber oversight should sit alongside financial and operational risk.

AI agents need a leash, not a blindfold

Then there is the newest and least understood risk: AI agents. Alex Yohn, who serves as CIO, Privacy Officer and Security Officer at the payroll company Wagepoint, warned about automated systems that get broad permissions and vague goals.

"Most fear around AI security centres on malice. Real failures are usually much duller," Yohn said. "Give an AI agent a goal and broad access, and it will find the easiest path to that goal, including paths nobody intended. It needs no bad intent, only weak boundaries."

His fix, he admits, is dull too. Give each agent only the access it needs. Log everything it does. Require a human to approve any action that moves money or data. Test the limits yourself before the agent discovers them. Yohn framed it as zero trust applied to a non-human actor: the risk is not a machine that wants to cause harm, but one that does its job better than the guardrails expected.

That insight matters because agents are being wired into real business processes fast, from customer service to finance. Every integration is a new set of permissions, and permissions that made sense for a human employee can be far too broad for software that never sleeps.

What a business can actually do this week

The advice from these four executives converges on a short, realistic to-do list. Decide which AI tools staff may use and block the rest. Check the privacy and data-retention settings on every approved tool. Review which systems have AI agent integrations and cut their permissions to the minimum. And get the board asking questions about cyber exposure on a regular schedule, not just after an incident.

None of this requires a new product or a bigger budget. Taken together, the message from Canada's cyber leadership is that AI may change the speed and scale of cyber risk, but the answer is still basic controls, clear rules and direct oversight. The companies that keep their data safe will be the ones that treat a copy-paste as seriously as a break-in.

Frequently Asked Questions

Why is pasting data into a free AI tool risky?

According to Xero's security chief Suzy Clarke, data you paste into a free AI tool may be stored or used to train the model, depending on your privacy settings. Once that happens, she says, it is very hard to get the data back. A client email or financial document shared this way can leave the company's control with no breach and no alarm.

Is AI really creating brand new cyberattacks?

Not necessarily, according to Kyndryl Canada's Denis Villeneuve. He says AI is not necessarily creating entirely new risks, but it can accelerate familiar ones by helping attackers find and exploit vulnerabilities faster. That shorter window between discovery and exploitation is what changes the calculation for security teams.

What should a company's board do about AI cyber risk?

BDO Canada's Rocco Galletto says boards should treat cybersecurity as a business risk and govern it like one. Directors do not need to become technical experts, but they should understand the company's exposure, set a clear risk tolerance, demand evidence that resilience measures work, and make sure security decisions align with business strategy.