How trust can drive competitive advantage in the age of AI

First comes the hype, then the disillusionment. As organizations embed AI into ever more business processes, trust in the technology appears to be stalling. One recent Gallup survey found that while more Americans than ever are familiar with the technology (70% versus 64% in 2024), the share who believe it does more harm than good rose to 39% this year. 

Concerned about the potential security, privacy, regulatory and reputational repercussions, stakeholders are demanding stronger governance and accountability.

Business and IT leaders should embrace these calls. Far from slowing down the pace of AI-driven innovation, good governance can actually accelerate it.

Why trust is waning

AI is well beyond the pilot stage. From customer-service chatbots to CV-screening HR assistants (HR), and code-writing DevOps apps, it’s used in a growing range of enterprise scenarios. But only 27% of Americans now say they trust businesses “a lot” or “some” to use AI responsibly, down from 31% in 2025. Why the decline?

Greater AI use means awareness is growing about its shortcomings. More people are seeing hallucinations and leaks, AI-powered fraud and cyber-attacks first hand. MIT’s AI Incident Tracker put incidents at an all-time high in 2025. 

AI agents add an extra layer of concern. They act autonomously at machine speed, with potential real-world consequences. If not even model makers can keep them from engaging in unexpected and malicious behavior, how can regular businesses feel confident?

Corporate policies are still nascent, with adoption often outrunning governance. Models and agents are frequently over permissioned, with accountability, auditability and human oversight often neglected.

Shadow AI compounds these risks. Incidents related to unmanaged AI rose from 20% of those recorded by IBM last year to 43% this year. And they are having an impact on the bottom line. AI-related data breaches grew to 21% this year from 13% last year. Data poisoning, model inversion and prompt injection were among the most common causes. Model inversion attacks cost on average $6.1 million, 18% higher than the global average.

Time for the board to take notice

If this doesn’t already make improving AI governance a board-level issue, regulation will. Authorities in different regions have thus far taken slightly different approaches. But with the exception of the US, the general trend is one of greater scrutiny of AI.

The EU’s AI Act stands out for being the first and arguably most comprehensive legislation. In the UK, regulators are taking a more industry-specific approach. But all frameworks focus on the importance of business resilience, as rule-makers fret over the potential damage that unmanaged, autonomous AI could do.

In response, organizations must find a way to formalize management of security and privacy risk, and enhance transparency, accountability and reliability. There’s a clear business advantage for doing so, argues Adam Ely, General Manager, AI Security at Check Point. “Organizations that can show they’re in control of their AI win the deals, pass the audits and earn the confidence to deploy more,” he says. “Those that can’t will find customers and regulators slowing them down.”

A study from GRC specialist IO shows clearly the direction of travel. It revealed that 28% of US/UK organizations required their suppliers to be compliant with ISO 42001, the AI management standard, last year versus only 2% in 2024. Standards like this help embed best practices and streamline compliance with regulatory requirements.

Start now for maximum advantage

IO CEO, Chris Newton-Smith, says the businesses that take a proactive approach to AI governance will be better placed to manage risk without putting the brakes on innovation.

“Rather than retrofitting controls around AI systems that have already been deployed without adequate guardrails, or trying to introduce governance after problems have emerged, organizations should build governance into their AI strategy from the outset,” he argues.

“This means establishing clear policies, accountability, oversight and risk controls before AI is deployed, while continuing to monitor and adapt those controls as the technology evolves. Done properly, governance can help organizations protect sensitive customer and business information, identify and mitigate emerging risks, and give employees clear parameters for using AI responsibly.”

Too often governance is viewed as a barrier to innovation. But if organizations already have a framework in place to deploy AI safely, they can “move faster when opportunities arise” rather than being forced to pause to retrospectively address risks, Newton-Smith continues.

“The result is a more sustainable approach to AI adoption – one that enables businesses to innovate at speed while keeping risk controlled, measurable and aligned with customer expectations,” he says.

Security as a trust signal

Cybersecurity must sit front and centre of any trust-building efforts, argues Check Point’s Ely. He explains that organizations must understand where they are using AI, protect the underlying data, manage identity and access securely, and install guardrails for runtime protection. Any strategy must be backed by “clear policies, a named owner for AI risk and regular reporting to the board”, he says.

IO’s Newton-Smith adds that information security, privacy and AI governance are fundamentally connected disciplines, and should be managed as such. “Organizations that take a centralized approach can map risk across these areas and address common risks accordingly,” he argues.

The benefits speak for themselves. 

A Deloitte study found that “high-trust” organizations were most likely to focus on cybersecurity, as well as data and governance. These same companies were also 18 percentage points more likely to be in the top third of organizations achieving their expected benefits from AI. Such benefits might include enhancements to products and services, improved innovation and growth, and greater cost-efficiency and productivity.

“Think of brakes on a car: they’re what let you drive fast. When security teams can see which AI tools are in use, control what data goes into them and monitor what agents do, the default answer to a new AI project becomes ‘yes, with guardrails’ instead of ‘not yet’,” concludes Check Point’s Ely.

“Building trust into AI from the start removes the friction that stalls projects in review, and lets businesses move from pilots to production much faster.”

As businesses brace for a new era of AI-driven competition, trust will increasingly be the difference between success and failure.