Beyond the Hype: A Strong Case for Human-Centric AI in Security

The security industry must cut through the AI hype and recognize why artificial intelligence in physical security requires human oversight to bridge the gap between marketing promises and operational reality. In response to Mirza Bahić’s article “AI in Security: Balancing Between Digital Savior and Snake Oil,” published in a&s Middle East, COM-SUR founder Gautam Goradia exposes the uncomfortable truths about false positives, missed threats, and the critical limitations that vendors rarely discuss.

By: Gautam D. Goradia
E-mail:gautam@comsur.biz

Gautam D. Goradia

After 11 years in the security industry, I’ve witnessed countless technology waves promising to revolutionize surveillance. But Mirza Bahić’s recent analysis strikes at the heart of what we practitioners know but rarely discuss publicly: AI in security is powerful, but it’s failing because we’ve forgotten the fundamental discipline of human oversight.

The failure of the AI-based gun detection system at Antioch High School in Tennessee isn’t an isolated incident—it’s symptomatic of an industry-wide problem. We’ve become so enamored with artificial intelligence that we’ve abandoned the basic principle that surveillance systems require human validation to be effective.

What the Experts Are Really Saying

When Hans Kahler from Eagle Eye Networks states, “Some vendors overpromise, suggesting AI can do everything. That’s hype, that’s not reality,” he’s voicing what many of us see daily. The gap between marketing promises and operational reality has never been wider. As he correctly notes, “AI will not — and should not — replace humans. It’s a powerful assistant, not a decision-maker.”

Sajjad Arshad from AxxonSoft Middle East captures this perfectly: “AI doesn’t drink coffee to stay more focused, but it also doesn’t understand intent.” While AI excels at scanning images and processing massive volumes of video without fatigue, “it falls short where it matters most — in understanding why something is happening.” His observation that “Security isn’t just about spotting anomalies. It’s about judgment” resonates deeply with my own experience.

This isn’t just theory. When Arshad mentions getting “truly bizarre requests from clients — the kind that make us stop and wonder: who told them these fairy tales in the first place?” he’s describing the real-world impact of AI overselling. Marketing, as he puts it, “tends to sell dreams, while AI delivers something a little messier.”

Mohammed Soliman from McLarty Associates puts it bluntly: “Marketing often exaggerates the promises of what AI in security can do… they pitch fully autonomous security or zero breaches, but that’s overselling it.” His analogy is spot-on: “AI is brilliant at shrinking the haystack — flagging 100 suspicious events out of a million — but it’s still up to humans to find the needle.”

Daily Human Oversight

What strikes me most about these expert observations is how they all point to the same fundamental gap: the absence of structured daily human review. We install cameras, deploy AI analytics, and then somehow expect the system to run itself. This is where we’ve fundamentally lost our way.

In my work developing COM-SUR, I’ve seen this pattern repeatedly. Organizations invest heavily in AI-powered security systems, but they fail to implement the basic discipline of daily footage auditing. Surveillance is installed but not audited. AI runs but is not validated. The result? Systems that generate noise rather than actionable intelligence.

The real breakthrough isn’t more sophisticated AI—it’s bringing back the discipline of systematic human oversight. This means implementing what I call the Daily CCTV Video Footage Auditing process, where operators can audit 24 hours of footage in minutes without missing critical events. It means creating structured workflows that allow humans to seamlessly review footage, tag exceptions, and build the ground-truth feedback that makes AI actually work.

Why COM-SUR

The security industry often sidesteps a tough reality: AI failures stem from deep-rooted challenges that persist even among top players. AI systems falter when trained on scarce or low-quality data, struggling to handle unique, site-specific scenarios. They frequently miss concealed or novel threats that don’t align with their training. Without contextual judgment, AI misinterprets complex situations, and its biases can spark misguided actions, undermining trust. Opaque decision-making processes further erode confidence, while fine-tuning models proves tough without consistent, high-quality feedback. Ultimately, AI’s effectiveness crumbles in isolation, demanding seamless integration to bridge these gaps.

COM-SUR addresses these challenges directly. Unlike traditional systems that require operators to sift through hours of footage, COM-SUR enables rapid, structured auditing, allowing users to review entire days of multi-camera footage in minutes. The system generates site-specific, timestamped data that AI vendors can train on, creating what is called “Better AI” through continuous human feedback.

But more importantly, COM-SUR creates Explainable AI workflows. When Saif AlRefai from OPSWAT notes that “AI systems can make accurate predictions, but they often operate as ‘black boxes,'” he’s identifying a critical vulnerability. COM-SUR builds transparent, documented evidence for every reviewed incident, eliminating the black-box problem.

The system works with all video sources—CCTV, drones, body-worn cameras, even smartphone footage—creating a unified auditing approach that scales across different technologies and environments. This isn’t just about better software; it’s about establishing a new operational discipline.

The Ministry of Education is introducing a COM-SUR-based curriculum in Grades 11 and 12.

The Government Recognition Factor

The validation of this approach goes beyond market acceptance. The Government of India has formally recognized “CCTV Video Footage Auditor” as a new skill under its Ministry of Skill Development. The Ministry of Education is introducing a COM-SUR-based curriculum in Grades 11 and 12. This institutional recognition reflects a growing understanding that video surveillance requires specific skills and systematic approaches to be effective.

As part of our National Service, we’re providing COM-SUR ‘Ultima’ (highest version) free (conditions apply) to Indian police, defense forces, and paramilitary units because national security demands the highest standards of surveillance discipline. When stakes are this high, there’s no room for unvalidated AI or undisciplined surveillance practices.

Our Social Commitment

As part of our societal purpose, we’re providing the COM-SUR ‘Business’ version free (conditions apply) to the following vulnerable targets: (1) All Places of Worship – Globally, (2) All Low-Budget Government Schools – Globally, (3) All Zoos – Globally. These are high-emotion, high-sensitivity spaces that are often under-resourced and overlooked. Our aim is to give back to society by offering an affordable, simple-to-use solution that ensures proactive surveillance, greater accountability, and far better outcomes than traditional methods.

“The interpretation of that data remains a human responsibility,”

Maher Yamout from Kaspersky

Beyond Technology: A Return to Fundamentals

The AI security industry has reached a tipping point. Over-promised solutions are being exposed through real-world failures, and the gap between marketing claims and operational reality is becoming impossible to ignore. As Maher Yamout from Kaspersky correctly observes, “The interpretation of that data remains a human responsibility.”

The answer isn’t abandoning AI—it’s grounding AI in proven operational disciplines. Cameras are sources of information that deliver optimal outcomes only when used correctly. Not everyone needs AI, but everyone needs structured human oversight, systematic data management, and standardized reporting.

COM-SUR represents this missing discipline—the structured, timestamped, explainable auditing process that transforms surveillance from reactive monitoring to proactive intelligence. It’s time to bring sanity back to security by embracing the essential truth that human intelligence must guide artificial intelligence, not the other way around. The mantra should be clear: “AI when you need it, human intelligence all the time”.