A major crisis is unfolding across corporate America as the widespread deployment of AI agents fails to deliver promised financial returns while tearing organizations apart from within, new research reveals. With 97% of business leaders having implemented the technology in the past year, fewer than 30% are seeing real benefits, creating a dangerous gap between massive investment and minimal return that executives warn could trigger significant operational corrections.

Two-Thirds of companies report security breaches

The security picture looks alarming. Two-thirds of executives believe their companies have already experienced data leaks or security breaches because workers used unapproved AI tools.

 More than one-third, 36 percent, have no formal plan for watching over their AI agents. Another 35 percent admitted they could not immediately shut down an AI agent if it went rogue. Thirty-five percent of employees have entered company secrets into public AI tools.

Deep Shah, a software engineer at Google, explained that “there are multiple challenges you will find when you try to deploy that system at scale.” He pointed to costs as the first major obstacle. Running AI agents requires constant spending, and poorly designed systems end up burning cash instead of saving it.

The problem goes beyond technical issues. Three-quarters of executives confessed their company’s AI strategy exists “more for show” than as actual guidance for employees.

Nearly half, 48 percent, called their AI adoption efforts a “massive disappointment.” Another 39 percent lack any formal plan to drive revenue from AI tools. The pressure has gotten so intense that 73 percent of chief executives report stress or anxiety about their company’s AI strategy, with 64 percent fearing they could lose their jobs if they fail to lead the transition.

Only 29 percent see real returns despite heavy use

Meanwhile, a separate analysis by Lyzr AI based on 200,000 user interactions, 3,000 demo requests, and 2,000 conversations with business and tech leaders found that 62 percent of companies exploring AI agents lack a clear starting point. Another 41 percent treat them as side projects. Thirty-two percent stall after pilot programs and never reach full operation.

Chris Han, who helps run ThinkingAI in China, said popular tools like OpenClaw cannot meet corporate needs. Business users need to figure out memory management, agent teams, and communications, tasks that the current tools handle poorly.

Only 29 percent of organizations report significant returns from generative AI, and just 23 percent from AI agents, even though 70 percent of employees and 94 percent of top leaders use AI tools for at least 30 minutes every day. Sixty-four percent of executives spend two hours or more with the technology daily.

If you're reading this, you’re already ahead. Stay there with our newsletter.