As Businesses Rely More on AI, Costly Errors and New Risks Are Emerging

Artificial intelligence has moved from experimentation to everyday business operations. Companies are using AI to write code, manage customer service, analyze data, automate accounting, and support decision-making. While many organizations continue reporting productivity gains, a growing number are discovering that AI can introduce new costs, unexpected errors, and operational risks that are proving harder to manage than early forecasts suggested.

Workers Are Saving Time, But Businesses Are Struggling To Capture The Value

The latest AI at Work report from Boston Consulting Group found that 74% of non-managerial white-collar workers now use AI regularly, up sharply from a year earlier. More than 40% reported saving at least one full workday per week through AI tools. Yet many companies have not figured out how to translate those time savings into measurable business gains. Nearly half of respondents said they spend more time directing and managing AI systems, creating what researchers described as a “joy paradox” where AI can simultaneously improve and complicate work.

Accounting Mistakes Show How Small Errors Can Grow

One area drawing particular attention is finance and accounting. Industry experts have warned that AI-generated bookkeeping and reporting errors can quietly spread through financial systems before being detected. Common problems include transaction misclassification, inaccurate tax treatment, duplicate entries, and AI-generated explanations that appear convincing but are incorrect. A single accounting error can affect financial reports, tax filings, and business decisions, creating hours of manual review and correction work that erodes the productivity gains AI was meant to deliver.

When AI Systems Act On Their Own

The risks become larger when AI moves beyond generating text and begins taking actions. In April 2026, software company PocketOS reported that an AI coding agent deleted its production database and backups in seconds after attempting to resolve a technical issue. According to the company, the AI accessed credentials, executed a destructive command, and disrupted customer reservations and records. Although the data was later restored, the incident highlighted concerns about granting autonomous AI systems direct access to critical business infrastructure.

Businesses Are Learning That AI Errors Can Become Expensive

Many organizations initially viewed AI as a path to lower labor costs. Recent evidence suggests the financial equation is more complicated. Gartner projects global IT spending will reach $6.31 trillion in 2026, driven partly by AI-related investments in infrastructure, cloud services, and software. An MIT Computer Science and Artificial Intelligence Laboratory study cited in industry reporting found that AI was cheaper than human labor in only 23% of examined computer-vision tasks, while implementation and maintenance costs exceeded labor costs in most other cases.

The Hidden Cost Of Tokens And Computing Power

Unlike employees whose salaries are relatively predictable, AI systems generate costs every time they process information. Companies increasingly pay based on “tokens,” the units used to measure AI usage. As businesses deploy AI agents capable of handling complex tasks, token consumption can rise dramatically. Industry analysts have reported cases where AI-related usage costs exceeded employee costs within months, prompting some firms to reconsider how broadly they deploy advanced models.

Real-World Lawsuits Are Testing AI Deployments

Businesses are also facing legal scrutiny when AI-powered systems fail to perform as expected. A Pizza Hut franchisee operating more than 100 restaurants filed a lawsuit alleging that an AI-assisted delivery management platform contributed to delays, declining customer satisfaction, and more than $100 million in damages. The franchisee claims the system created operational breakdowns after giving delivery drivers new visibility into restaurant workflows. Pizza Hut has said it is reviewing the allegations and will respond through legal channels.

Data Quality And Governance Are Becoming Competitive Advantages

Experts increasingly emphasize that AI systems are only as reliable as the data they receive. Poor-quality, fragmented, or outdated information can produce flawed outputs regardless of how advanced the model is. Businesses also face challenges integrating AI into legacy systems, maintaining compliance with privacy regulations, and ensuring employees have the expertise needed to review AI-generated work. As adoption accelerates, governance frameworks are becoming as important as the technology itself.

Companies Are Adjusting Their AI Strategies

Rather than abandoning AI, many organizations are refining how they use it. Some are shifting to smaller, specialized models that cost less than large general-purpose systems. Others are limiting AI to narrow tasks such as data processing, draft generation, or routine categorization while keeping humans responsible for final decisions. Businesses are also investing more heavily in oversight, auditing, and review procedures designed to catch mistakes before they affect customers or operations.

The Next Phase Of AI May Be Defined By Control, Not Adoption

The first wave of corporate AI adoption focused on how quickly organizations could deploy the technology. The next phase may focus on how effectively they can manage it. Companies are discovering that successful AI implementation requires more than access to powerful models. It demands clear governance, high-quality data, human oversight, and careful cost management. As AI becomes a standard business tool, the organizations that benefit most may be those that treat it not as a replacement for judgment, but as a system that requires constant supervision and accountability.