Employers, Employees Both Play Roles in Managing AI

Leveraging artificial intelligence—and protecting against its dangers—should be a responsibility shared by employers and employees, according to new research from the Transamerica Institute and Marsh McLennan Agency.

The Transamerica Institute, in its “Employers, Workers, and the New World of Work” report, found that 82% of employers are currently using or planning to use artificial intelligence to “augment their human workforce,” including 97% of large companies (with at least 500 employees), 95% of medium companies (between 100 and 499 employees), and 79% of small companies (fewer than 100 employees).

Among surveyed employers currently using AI to supplement human judgment—or planning to do so—85% of representatives from large companies reported plans to expand their use of the technology over the next three years, compared with 71% of medium companies and 51% of small companies.

A recent report by Marsh McLennan Agency, “2026 People Risk,” cautioned that employers should shift toward viewing AI as more of an augmentation tool, rather than a means of displacing or reducing their human workforce. MMA suggested “combining AI’s speed, scale and precision with human judgment, creativity and empathy.”

The Transamerica Institute survey found employers’ predictions of AI’s impact on the workforce loomed large: 89% of respondents representing employers currently using or planning to use AI said there will be implications for their company’s workforce, including that certain jobs will transform (67%), new jobs will be created (50%) and certain jobs will be eliminated (36%).

But employers should beware: They might need their workforce to harness the best of AI—and to protect against the worst of it, MMA’s report suggested.

Leveraging Employees to Transform the Business

“Most organizations still expect employees to integrate AI into their existing workflows rather than reimagining the work itself,” MMA’s “People Risk” report stated. “Without a clear strategic vision, work redesign, an AI-ready workforce and governance that can keep up with the pace of change, AI will remain on the periphery: improving individual parts of the business without transforming the whole.”

MMA’s report also revealed that human resources and risk professionals ranked “overcoming mindset-related barriers to AI adoption” sixth among AI workforce risks globally—while C-suite survey respondents ranked it as third. Among all respondents, top concerns included investment in AI and adoption without appropriate employee training and upskilling (40%); lack of AI knowledge in HR, limiting the “transformation of people practices” (38%); limited technology team capacity to support AI implementation, adoption and upskilling (38%); limited knowledge or awareness of AI risks and mitigations (38%); and employee resistance to AI due to fear of job loss or change fatigue (37%).

The MMA report also concluded that uncertainty about whether employees are being properly equipped to use AI highlights a “growing mismatch between AI spend and AI capability.”

Employees themselves are aware of the need to invest in their own abilities, especially in a rapidly changing labor market: MMA’s “Global Talent Trends 2026,” a separate report, found that 63% of employees said they would trade a 10% pay raise for better AI and digital reskilling opportunities.

Protecting Against AI, Digital Threats

HR and risk professionals did express support for equipping employees with the AI tools they need—specifically to prevent AI-related cybersecurity breaches, according to the “People Risk” report. Respondents ranked as their top “AI priority” the need to train employees to identify malicious, AI-generated misinformation or disinformation (41%)—reflecting a need to build AI and cybersecurity readiness across all levels of the workforce.

Respondents to MMA’s survey ranked “inadequate cyber threat literacy” as their top “people risk.” Technology skill shortages, including cyber skills, ranked third.

More than 70% of surveyed organizations reported experiencing at least one material third-party cybersecurity incident in the past year, according to the MMA report. Phishing and social engineering were the most common entry points, with compromised log-in details a leading cause of ransomware and data breaches. In addition, as AI tools have become better at mimicking reality, more convincing deepfakes have become more difficult to identify.

“To help employees embrace this technology with less fear, it is essential for organizations to communicate their vision for how they will integrate AI into their corporate strategy,” the MMA report stated. “They should involve workers at all levels on AI design and implementation, while providing ongoing reskilling for the future of work. Crucially, employers should provide a ‘safe space’ for employees to train, test and master these tools, while ensuring privacy and security of information.”

Best Practices

Marsh McLennan recommended that HR and risk management professionals treat digital literacy, judgment and accountability as core risk controls, supported by shared ownership between risk, HR and technology departments.

To manage cybersecurity risks, including those inherent in AI, Marsh McLennan recommended employers also:

• “Reframe cyber risk to consider broader digital and technology risk;
• Identify, assess and recruit talent with strong cybersecurity skills;
• Equip employees and providers to maintain vigilance and operate safely under pressure;
• Expand cyber scenarios within risk planning;
• Treat workforce behavior, training and culture as first-line cyber controls; and
• Strengthen security through robust controls while reducing high-risk exposures.”

Transamerica surveyed 1,900 U.S. for-profit employers in late 2025.

Marsh McLennan fielded its 2026 National Benefits Strategy Survey among 616 employers in January and February. The company collected responses for its “People Risk 2026” report from 2,256 human resources professionals and 2,261 risk professionals, including 1,000 based in the U.S., between October and November 2025.