AI adoption is splitting companies as leaders push for faster use

Companies are pushing employees to use AI faster than they are building the rules, training and workflow redesign needed to make it work, and that mismatch is now showing up as division inside the workplace. In a March 18, 2025 survey of 1,600 U.S. executives and knowledge workers, 68% of C-suite leaders said AI adoption had caused division at their company, while 42% said it was tearing the company apart.

The friction is not just cultural. The same survey found tension between IT teams and other business lines at 68%, and between executives and employees at 63%. Even as 73% of companies said they were investing at least $1 million a year in generative AI, only around one-third had seen significant return on that spending. The result is a familiar corporate pattern: expensive software, unclear rules and uneven uptake.

That gap is widening because workers are being asked to adapt before firms have explained what success looks like. The survey found 31% of employees admitted they were sabotaging their company’s AI strategy, including by refusing to use the tools. Around one-third said they were paying out of pocket for generative AI tools because their employer did not provide the ones they wanted, a sign that adoption is happening in a fragmented, unofficial way rather than through a controlled rollout.

The problem is showing up in usage data as well. In BCG’s June 26, 2025 AI at Work survey, regular AI use among frontline employees stalled at 51%, even as more than three-quarters of leaders and managers said they used generative AI several times a week. BCG found that employee sentiment changed sharply when leadership support was strong: the share of workers who felt positive about GenAI rose from 15% to 55%. Regular use also rose when employees received at least five hours of training and in-person coaching, underscoring that adoption depends on instruction, not just access.

Workplaces are still catching up. The American Management Association surveyed 1,252 North American knowledge workers and found 58% believed they were behind in keeping up with AI. A September 2025 Harris Poll for MasterClass showed 49% of workers felt direct pressure to adopt AI, while 55% said they did not know where to start and 66% said they had to teach themselves on the job. Half-trained employees, vague expectations and weak governance create a mix that can hurt productivity, strain morale and expose companies to legal and compliance risk if AI use drifts outside policy.

By 2026, that tension had hardened into resistance. Writer’s 2026 AI Adoption in the Enterprise report, based on 2,400 knowledge workers across Europe and North America, found employees avoiding AI tools, ignoring new processes or pretending to use systems without actually changing how they work. The message from the data is blunt: companies that demand AI adoption first and figure out the operating model later are not accelerating transformation, they are manufacturing confusion.