Artificial intelligence adoption is accelerating across industries. Since the introduction of generative AI tools a few years ago, many organizations are experimenting with AI-driven capabilities.
According to the Genetec 2026 State of Physical Security report, AI is now a top priority for security departments. AI ranked alongside access control and video surveillance as a key focus area for 2026. At the same time, 70% of end users report concerns about how AI systems are designed and implemented, particularly around auditability, bias, and data use.
The rhetoric around AI today includes a lot of speculative ideas that aren’t yet road-tested or reliable. It’s true that AI is reshaping physical security, but systems integrators can help their customers think through responsible applications, realistic goals and discern what is true and what is hype.
How AI is really used today
It is important to realize that AI is a technology, a set of tools. But much more important than the tool is the outcomes manufacturers and integrators can deliver with it.
To the operator, the AI tools are important because they improve the speed at which operators are able to complete their tasks. AI has been present for years in automatic license plate recognition (ALPR) and video analytics, enabling search for things like “a blue car” or “a person wearing a hat.”
What’s different now is the addition of large language models (LLMs), which make the AI layer more visible and dynamic. LLM interfaces can significantly reduce training time. Instead of thumbing through a manual to find the answer to a question, operators can type a query in plain language, and the system will quickly find the data.
For example, if an operator or investigator needs footage of a person spotted outside in the early morning hours, they can use natural language prompts to search multiple cameras and hours of footage. They can input a simple phrase such as, “a man wearing a yellow hat and a black jacket this morning” or “a person with a backpack” and the AI-enabled tools quickly pull footage that matches that criteria.
As relevant footage is identified, the operator validates and refines findings. The system uses AI to help identify similar footage of people, vehicles, and objects across scenes to support a summary of events.
This approach not only helps the user find what they’re looking for more quickly but also reduces cognitive load, enabling investigation teams to work more efficiently. Instead of scouring hours of footage and piecing together various clips, they can easily present a coherent storyboard that shows how events connect.
Assisting Human Operators, Not Replacing Them
Whether supporting investigations or day-to-day operations, AI is best used to help humans work more efficiently. AI systems do not possess ethical reasoning or contextual judgement, so they cannot replace human operators.
AI is never 100% reliable. AI-powered tools rely heavily on the quality of the data they receive. Poor data quality yields unreliable results. Unintended biases tend to creep in, distorting AI reasoning. Likewise, data sets can include errors and omissions that an experienced human would recognize but AI might not.
Critical thinking skills are key, and humans are responsible for making the decisions and accountable for the results. We know AI can support humans to make better decisions, but human oversight is essential to ensure AI tools don’t misinterpret or misrepresent the facts.
Implementing AI Responsibly
As AI use expands, so do the potential negative consequences of unmanaged risks. When considering AI-based solutions from different manufacturers, there are three guiding principles for the responsible use of AI:
- Privacy and data governance: The solution uses only datasets that comply with local data protection regulations and treats them with the utmost care, limiting access as needed.
- Trustworthiness and safety: The solution minimizes bias in AI models, rigorously tests them, and strives to always make them explainable.
- Humans in the loop: AI models do not make critical decisions by themselves. Humans are always in the loop and have the final say.
Manufacturers vary in their approach to responsible AI practices, so understand their posture on this. Also, work with your customers to think through where their data is stored, who has access to sensitive information, and who’s accountable for protecting that data.
AI Evolution’s Impact
AI is here to stay and is a powerful resource. New capabilities will emerge as machine learning models improve and more data is available. For systems integrators, the opportunity lies not in chasing every new feature, but in understanding where these tools genuinely add value. System integrators can help manage customer expectations about AI and separate realistic capabilities from unrealistic expectations.
By focusing on practical use cases, integrators can help their customers adopt AI in ways that are reliable, secure, and operationally meaningful. In the end, successful deployments depend less on the technology itself and more on thoughtful implementation and responsible use.
Alex Halliday is director of channel enablement, North America, sales, at Genetec.
