AI is forcing its way into life and work at breakneck speed. According to Stanford University’s 2026 AI Index, generative AI reached 53% population-level adoption in just three years, faster than the PC or the internet.
AI can transform how organisations work and help them achieve things that seemed impossible just a few years ago. It’s no surprise that they want to make the most of the opportunities. But as adoption accelerates, an important question arises. Are organisations putting the right safeguards in place to keep the risks under control?
Governance lags behind adoption
AI agents make this challenge more urgent. These systems go beyond responding to prompts. They can carry out tasks on their own, often with little human input. While the productivity possibilities are exciting, there are understandable questions about how much freedom organisations should give them.
According to Deloitte’s April 2026 research, only 21% of surveyed organisations had a mature governance model for agentic AI, even though 74% expected to use AI agents at least moderately by 2027.
That leaves a worrying gap between ambition and governance. Organisations need to decide what their AI agents can do independently and when a person must step in to approve an action. They also need to monitor how these systems behave and make sure someone takes responsibility for their decisions. Without that oversight, you risk handing increasingly important tasks to technology few people fully understand.
When AI gets things wrong
The consequences of AI mistakes can be serious. Sometimes, they can be life-altering.
In October 2026, Open Rights Group investigated the Home Office’s use of AI to summarise asylum interviews. It found that 22% of summaries failed initial expert quality checks, compared with the 9% the Home Office previously quoted. Inaccurate summaries could leave decision-makers with a misleading account of someone’s case.
AI can also influence how people think. An experiment cited in the February 2026 International AI Safety Report found that interactions with AI changed participants’ beliefs about trivia answers by 17 percentage points, compared with nine percentage points when they interacted with other humans.
These findings raise important questions about human oversight. Organisations need to make sure employees:
- Check AI-generated information against reliable sources before making important decisions.
- Feel confident challenging AI outputs rather than accepting them without question.
Security and business risks are growing
Evidence of AI-related harm in the real world continues to mount. Stanford University’s 2026 AI Index found that documented incidents involving AI rose by 55% in a single year, from 233 to 362 cases. As organisations rely more heavily on AI, they need to understand where these systems might fail and what damage those failures could cause.
Security presents another major concern. IBM’s 2025 Cost of a Data Breach Report found that 13% of organisations studied had suffered breaches involving their AI models or applications. Of those affected, 97% lacked proper AI access controls.
These figures highlight the dangers of giving AI systems access to sensitive information without adequate protection. As organisations connect AI to more of their existing technology, they need to control which systems can access their data and who can use them. Otherwise, the technology they hope will improve their operations could introduce serious new vulnerabilities.
Building confidence into AI adoption
AI offers organisations enormous opportunities, but they need to prepare properly before putting it to work. That means considering governance requirements when planning technology investments and making sure the right safeguards are in place from the start. Getting these foundations right can help organisations embrace AI with greater confidence.
Lombard Technology can support clients early in this process, helping them understand their technology and infrastructure requirements while exploring how to fund their AI ambitions responsibly.
__
About Sam
Sam Scott is a technology specialist at Rigby Capital. He works across Rigby Capital and Lombard Technology, helping clients better understand technology and how it links to their business ambitions. Sam’s career spans healthcare, medtech and the wider IT sector. Along the way, he’s built experience in infrastructure, software, cybersecurity and AI. As technology evolves at pace, it’s this knowledge that helps Sam show businesses what matters and where to go next.
