The hidden AI risk companies can’t afford to ignore

OneTrust's Corey Rice warns that weak AI governance is causing costly delays, reputational damage and regulatory risk for businesses racing to adopt AI. Organisations often introduce governance too late, rely on manual checks, and fail to assign clear responsibility, leading to stalled projects, biased outcomes, and loss of trust.

Opinion: As businesses race to adopt AI, weak governance is creating costly delays, reputational damage and regulatory risk, argues OneTrust’s Corey Rice. Artificial intelligence is quickly becoming embedded https://www.forbes.com.au/news/innovation/albanese-creates-office-of-ai-so-australia-can-be-more-than-a-data-warehouse/ in modern business, promising greater efficiency, faster decision-making and new opportunities for innovation. Yet many organisations still treat governance as an afterthought – something to address after deployment rather than a capability that should shape AI projects from the start. In the rush to innovate https://www.forbes.com.au/news/innovation/faculty-the-uks-billion-dollar-ai-training-program-launches-in-australia/ , governance is too often seen as a compliance exercise or operational hurdle that slows progress. In reality, organisations without effective governance increasingly face stalled projects, costly rework, reputational exposure and rising regulatory risk. Without a shift in approach, governance gaps could become one of the biggest barriers to AI-driven growth. Errors and missteps A common mistake is introducing governance too late. Organisations write policies, adopt frameworks and define principles, but rarely connect these controls to the systems where AI operates. As a result, organisations are relying on manual checks and periodic compliance reviews to oversee AI systems that can change rapidly. This does not reflect how AI works in practice, as models and the data they use can shift over time. The way organisations assign governance responsibilities often makes these challenges worse. Legal, compliance, cyber security, risk, engineering and data teams frequently share responsibility, but no single group oversees the whole system. That makes it harder to spot problems early and respond quickly when they arise. These gaps are becoming more visible in practice. When organisations identify governance issues too late, they delay, rework or sometimes abandon products and services built on AI stacks — the technologies, frameworks and infrastructure that support AI systems. In some cases, organisations have used customer data to train models without proper consent. In others, they have embedded inaccurate, incomplete or biased datasets in production systems, creating risks that become far harder and more expensive to fix after deployment. Organisations also often fail to understand or monitor AI model behaviour early enough. This creates particular risks in high-stakes areas such as lending, recruitment and biometric surveillance, where automated decision-making systems can introduce bias. The results can cause real-world harm. Think unfair loan rejections or the targeting of specific demographic groups. The reputational cost One of the most immediate impacts of this approach is what is sometimes referred to as “compliance latency”, where initiatives are delayed or blocked because risks are identified late in the piece. The resulting rework can increase costs, delay launches and, ultimately, reduce revenue. Over time, these delays can become a clear competitive disadvantage. In many sectors, speed is critical. While one organisation is pausing to address governance issues, competitors may already be launching products, acquiring customers and strengthening their market position. Reputational damage may ultimately prove even more significant than any financial or regulatory penalties. As scrutiny from customers, employees, investors and regulators increases, organisations are being judged not only on what their AI systems deliver, but on how responsibly they are built and deployed. Biased results, misuse of customer data and decisions that are difficult to explain can quickly undermine trust and damage a company’s reputation. With trust in AI still forming, even a single misstep can have long-lasting consequences for credibility. How to address the risks To address these challenges, organisations need to build governance into AI projects from the beginning rather than treating it as a final check. This means checking that data is accurate, appropriate and used with proper consent, while also testing models before launch for risks such as bias or unreliable behaviour. Just as importantly, organisations need to build these rules into the AI systems themselves, so safeguards are applied automatically rather than relying on people to carry out manual checks. With ongoing oversight in place, businesses can move beyond small-scale experiments and use AI more confidently across the organisation. Organisations should treat governance not as a brake on innovation, but as a way to adopt AI more quickly and safely. Corey Rice is the director of sales engineering and implementation at OneTrust . Want to see more Forbes articles on your feed? Tap here to make Forbes Australia a preferred source on Google. Look back on the week that was with hand-picked articles from Australia and around the world. Sign up to the Forbes Australia newsletter here or become a member here .