Six in Ten Large Companies Have Slowed or Paused AI Deployments

Six in ten large companies have slowed, paused or pulled back a planned artificial intelligence deployment in the past year. The reason is not the technology. It is trust.

That is the central finding of a new report from FTI Consulting, titled "AI's Second Act, Where Safety, Risk and Trust Collide" and released September 29. The firm polled 1,600 senior business decision-makers across seven markets: the UK, the US, Germany, France, Ireland, Belgium and Spain.

Cybersecurity emerged as the top concern. And 81 percent of respondents said unclear or shifting AI rules have already created material problems for their businesses, from higher risk aversion to direct pressure to slow adoption.

"AI has moved from hype to hard reality," said Jon Priestley, a senior managing director in FTI Consulting's strategic communications segment. "Companies still see enormous opportunity, but enterprise AI adoption is now being shaped by cybersecurity risk, regulatory uncertainty and a growing trust gap inside organisations."

Priestley said companies are "entering a more cautious phase." The next stage, which he called "AI's second act," will be defined not by who moves fastest, but by who "can deploy it responsibly." The Times covered the findings this week, reporting that more than half of companies have paused or slowed AI adoption over fears about whether they can trust the technology.

Cybersecurity Tops the Worry List

The survey lands as security teams confront a new class of AI-specific threats. FTI's data points to shadow AI, employees using unapproved tools, and governance gaps as forces reshaping deployment timelines. The report argues that businesses are not walking away from AI. They are layering risk management and governance onto adoption plans, and internal confidence now sets the pace.

The pattern is consistent across the seven markets surveyed. Reputational risk, regulatory exposure and doubts about model reliability were cited together as the brake on rollouts, a shift from the speed-first deployments that defined the first wave of enterprise AI.

The McDonald's Test Case

The trust question is already playing out in court. On October 2, a proposed class action was filed in federal court in Chicago accusing McDonald's of using an AI-powered pricing system to coordinate menu prices across its franchises and company-owned restaurants, in alleged violation of federal antitrust law, Reuters reported.

At issue is a machine-learning tool the company says it has used for more than a decade. The system analyzes millions of daily transactions across nearly 14,000 US restaurants to recommend what executives call the optimal price for each menu item. The gaps it can produce are stark. At one McDonald's in Fresno, California, a Big Mac costs $5.69, according to the company's mobile app. At a location two miles away, the same burger costs $6.89, the New York Post reported, citing Reuters.

The complaint, filed on behalf of Illinois customer Michael Thomas, alleges the platform lets franchisees share nonpublic sales data through a centralized system, undermining competition among independent operators who are supposed to set prices on their own. McDonald's called the allegations speculative and said it will defend itself vigorously. "AI does not set the price of a Big Mac or any other menu item," the company said in a statement. Franchisees, it insists, make the final pricing decisions themselves.

FTI flagged the expansion of the McDonald's pricing tool in its own news roundup this month as a case raising fresh questions around consumer trust and regulation.

Regulation Closes In

The regulatory backdrop is hardening. The FTI report was timed ahead of Dublin's International AI Summit, where Priestley said businesses will look for practical guidance on cyber resilience, governance, disclosure and liability. For the 81 percent of executives who say AI rules have already caused material business problems, that guidance cannot come soon enough.

The warnings are getting louder beyond the boardroom. Anthropic recently cautioned that advanced AI models could pose catastrophic risks, and Microsoft AI chief Mustafa Suleyman likened AI to "a new silicon species," both developments cited in the Times coverage of the FTI findings.

For now, the data points to a deliberate slowdown rather than a retreat. Companies still see the opportunity. They have simply stopped racing to deploy at all costs. What replaces the race, FTI argues, is governance. The winners of AI's second act will be the companies that earn trust first and scale second.