At the conference of the world’s largest HR professional association, AI governance was a big concern, according to an attendee
While artificial intelligence (AI) now shapes hiring, performance reviews and workforce planning inside many organizations, research from the Society for Human Resource Management (SHRM), the world's largest HR professional association, suggests that fewer than half of all organizations have any AI policy in place.
The “guardrails-be-gone” finding comes from SHRM’s survey and report titled Navigating AI in the Workplace: 2026, which surveyed nearly 5,900 U.S. employees The report was a centerpiece unveiling at SHRM’s recent annual conference in Orlando, Fla.
The report finds that where a workplace AI policy exists, nearly one-third (30 per cent) of employees have breached it, and among those, another 31 per cent said they would contravene it again in an effort to save time. Results suggest that the gap between adoption and governance is not shrinking. It is widening.
Leaders are all in, the workforce is not
Executives and management have embraced AI well ahead of the people who use it every day, presenting a gap.
According to the report, 63 per cent of employees say their chief executive officer shows real interest in AI, and 60 per cent of directors now raise it in performance conversations. Yet fewer than half of employees use AI in their own work. Enthusiasm at the top has not yet translated into engagement on the ground.
The same gap shows up in employer-reported data too. A conference session hosted by Silberman Law, a boutique U.S. employment law firm focused on AI in the workplace, reported that 62 per cent of organizations use AI in some form, including 43 per cent in human resources and 27 per cent in recruiting. One of Silberman’s most-meaningful statistics is that 89 per cent of organizations that use AI in HR said that it increased their efficiency.
“Due to that high favourable reporting, it makes me believe that we are just going to continue to see these numbers increase and we are going to see a continued increase in these different types of tools,” said Krystal Wellon, Lead Statistician at Silberman Law. “So, it's really no longer about ‘Is AI coming to HR?’ It's here. And we need to think about how we use it and how we get it right.”
AI tools are also changing the pace of work. The SHRM report found that 54 per cent of employees report a faster-paced workplace, and 63 per cent report a higher volume of work. But faster and busier does not guarantee better, creating a retention risk employers cannot ignore.
The liability hiding in the hiring stack
Here in Canada, AI utilization in hiring and performance decisions must still comply with Canadian human rights law, privacy law, and now Ontario's new disclosure rules. As of Jan. 1, 2026, amendments to Ontario's Employment Standards Act, 2000 (ESA) require employers to disclose when they use AI to screen, assess, or select job applicants. The rule arrived before most employers even built the tools or governance to comply.
Wellon also described an alarming case study in which an AI matching tool rated 97 per cent of all candidates as a low match overall, with only 0.25 per cent rated a good fit. When cross-checked against actual recruiter decisions, 70 per cent of candidates who were interviewed had fallen into the tool's low-match categories, and the tool separately rated 25 per cent of candidates who did not meet basic qualifications as a good match. That mismatch is the blind spot employers need to close before relying on these tools.
“So, the big takeaway here is that just because you've implemented an AI tool, you are not successfully using AI just because you've done that,” Wellon warned. “It's really more about making sure the AI is aligning with your objectives, and that you have the appropriate oversight involved.”
In Ontario, under section 8.4 of the ESA, employers that use AI to screen, assess or select applicants must disclose that in job postings. Ontario Regulation 476/24 sets the threshold at 25 or more employees, and section 15(7.1) of the ESA requires those records to be kept for three years. The rule does not enforce full transparency. Employers must say that they use AI, but not how. A single line in a job posting satisfies the requirement while telling applicants almost nothing about what the tool actually does or how it reaches a decision.
Ontario is legislating transparency in this area unlike the U.S., arguably putting Canadian employers ahead of the trend instead of reacting to it. The federal picture remains unsettled. Canada's proposed federal AI law, known as the Artificial Intelligence and Data Act, died when Parliament was suspended in January 2025, so the provinces are writing the rules first.
AI hiring best practices
AI can improve hiring, but only when it supports a disciplined process without replacing reasoned judgment.
The problem it can help solve is well documented. A Leadership IQ study, cited in another SHRM conference session hosted by Aesop Partners, an HR consulting firm, found that a staggering 82 per cent of new hires fail to meet expectations within their first year. It seems the fix is not more unbridled technology on its own. It is structured interviews built on a clear definition of the role followed by guard-railed AI used to keep the process consistent and compliant across all candidates.
The results can be significant. Aesop reported attrition down significantly at a Fortune 500 company and a retention improvement of 22 per cent at an automotive services firm.
In short, the automotive example illustrated that AI‑enabled, highly-structured interviewing does not just sound good in theory. It yielded a real‑world test where the structured group outperformed the traditional process and a provable return on investment.
Canada also sets a floor here that our southern neighbour does not. Under section 12.1 of Quebec's Law 25 (the Act respecting the protection of personal information in the private sector), a person must be told when a decision about them rests entirely on automated processing, and that person can request that the decision be reviewed by a human. Keeping a person in the loop is a legal requirement in Quebec, not simply good practice.
Employees are doing the math
Employees are already doing the math on what AI means for their own careers and do not appear to be landing confident answers.
The SHRM report uncovered that fewer than half of employees believe AI will improve their career prospects or job security. At the same time, 69 per cent believe that heavier users of these tools are likely to earn more, and 72 per cent believe that upskilling will raise their own wages. Employees see winners and losers taking shape thanks to AI, even if they are not certain which group they will land in.
Ted Flett is an employment lawyer and a partner at Zubas Flett Liberatore LLP in Toronto. He gratefully acknowledges assistance from Gideon Ampofo in completing this article.
