The integration of artificial intelligence (AI) in workplaces has increased steadily, with 72.9% of organisations reportedly implementing some form of AI. A recent study indicates that deeper integration is associated with increased risks rather than reduced ones.
Organisations with more extensive AI adoption are reported to be 40% more likely to encounter AI-related issues. This has been linked to what is described as a “visibility gap”, where increased use of AI in operations makes monitoring and governance more difficult.
Around 22% of organisations have experienced incidents that have led to unanticipated costs or security breaches. In addition, 59.7% consider such incidents to be likely in the future, indicating ongoing concerns around AI governance.
AI is increasingly embedded across developer tools, productivity software, and automated agents. Its widespread use is contributing to a growing focus on governance and oversight approaches.
When examining AI-related priorities for the coming year, IT leaders highlighted the following:
- Automating IT operations: 44.4%
- Deploying productivity tools: 41.0%
- Establishing AI governance: 36.7%
The focus on both productivity and governance suggests an effort to balance AI implementation with oversight and compliance requirements.
As organisations continue to adopt AI, several challenges remain:
- Unmonitored AI tools: Employees may independently use AI solutions, contributing to “shadow AI” outside formal oversight.
- Advanced AI capabilities: Some tools now include command-line functions and complex models that may not be fully captured by traditional monitoring systems.
- Vendor proliferation: The rapid addition of AI features across software products increases the number of tools requiring evaluation and oversight.
- Cost considerations: Subscription-based and variable pricing models can reduce financial transparency and complicate cost management.
Overall, organisations are increasingly focusing on how governance can evolve alongside AI adoption. This includes improving understanding of how AI systems operate, what data they access, and where potential risks may arise, with the aim of supporting more controlled and secure implementation.
