Enterprise leaders have poured significant investment into AI over the past several years. The returns, so far, have been inconsistent.
A recent Forbes Technology Council piece from ISACA’s Chief Global Strategy Officer makes this tension concrete. According to ISACA’s 2026 AI Pulse Poll of more than 3,400 digital trust professionals, only 22% of global respondents said AI ROI has met or exceeded expectations. The majority either reported disappointing returns, said it was too early to tell, or simply didn’t know.
That finding points to something important: AI performance alone does not determine ROI. The gap between aspiration and outcome often comes down to how AI is implemented, governed, monitored, and integrated into the organization.
The governance gap
The Forbes piece identifies several interconnected causes behind underperforming AI investments. Organizations often deploy AI into existing processes rather than redesigning those processes for AI. Data infrastructure is not prepared for AI processing. Employees are not adequately trained to work alongside AI systems. And without clear performance measurement, there is no reliable way to know whether an AI initiative is working or drifting off course.
There is also a deeper challenge around complexity. As AI models become more sophisticated, they can behave in ways that are difficult to explain, even for the people who built them. Without structured oversight, that opacity introduces risk that organizations may not fully recognize until something goes wrong.
The article argues that governance should be understood as an enabler of AI performance, not a constraint on it. Organizations that build the right structures around their AI deployments, clear alignment with business objectives, standardized processes, and ongoing performance tracking, are the ones positioned to realize the returns they are looking for.
Why this matters in physical security
For organizations deploying AI in physical security environments, the governance question is not abstract. It has direct operational consequences.
An AI video system that generates false positives at a high rate does not just create noise. It trains security teams to stop trusting the alerts, which effectively negates the value of the system. A deployment that has not been calibrated to the specific environment it operates in will underperform from day one, regardless of how capable the underlying technology is. And without clear accountability structures around who reviews alerts, who acts on them, and how decisions are tracked, it becomes impossible to measure whether the system is actually improving security outcomes.
These are exactly the kinds of governance failures the Forbes piece describes in broader enterprise AI contexts. They are just as present, and just as consequential, in security deployments.
How Koshee approaches this
The principles the Forbes piece lays out, strategic alignment, structured implementation, ongoing performance measurement, and human accountability, are built into how Koshee Protect is designed and deployed.
Our models learn the specific environment they operate in. Detection keeps humans in the loop, allowing teams to review what the system flagged before determining the appropriate response. And because Koshee operates within your own infrastructure, your organization retains control over its video data.
AI delivers on its promise when the governance around it is as carefully considered as the technology itself. For security teams, that means deploying systems that are accurate, explainable, and designed from the start to keep humans responsible for the decisions that matter.
Read the full Forbes article: Remedying Mixed Results on AI ROI With Structured Governance
Learn how Koshee builds AI video intelligence with accountability and transparency at the core: Contact Us here