Last month, four thieves executed one of the most audacious museum robberies in modern history. Two men in yellow safety vests used a truck-mounted lift to reach the balcony of the Louvre’s Apollo Gallery at 9:30 a.m., cut through a window with a power tool, smashed display cases and escaped on motorized scooters by 9:38 a.m.
In just seven minutes, the thieves stole more than $100 million in historic jewelry, including an emerald necklace gifted by Napoleon Bonaparte. In the process, they exposed a global truth: the world’s most visited museum could only record, not prevent, the crime.
It would seem logical that the yellow vests and an oddly placed lift would have registered some level of alarm. Unfortunately, the museum’s historical treasures were guarded by outdated surveillance methods, including a paltry number of obsolete cameras.
Post-incident investigation revealed that 75% of the Richelieu wing had no active CCTV coverage. And nearly two-thirds of the Sully wing, which houses priceless French paintings, had no cameras at all. The only outdoor camera near the point of entry faced the wrong direction.
But would more cameras have stopped this robbery? Probably not. For highly trafficked areas like museums, shopping malls, and transportation hubs, the problem isn’t a shortage of cameras. It’s a lack of intelligence.
Traditional video surveillance depends on human operators watching dozens of feeds in real time. Fatigue in this case can set in within minutes, leaving events unnoticed until after they occur. Footage after an event serves only as evidence, not as prevention, becoming reactive rather than proactive.
From Watching to Understanding and Preventing
Artificial intelligence changes this model entirely.
Instead of relying on motion detection or fixed rules, AI surveillance can continuously learn what “normal” activity looks like in each camera view, including pedestrian and customer traffic and behaviors, staff routines, deliveries and cleaning schedules.
Once the baseline is learned, the system flags anything abnormal.
In the case of the Louvre, an AI-enabled surveillance system might have noticed a truck pulling up in a restricted perimeter zone, the two individuals wearing masks and climbing on a lift, or a power tool being used against gallery glass.
These are not static rules. They’re behavioral anomalies. An AI-enabled system would have sent notifications minutes earlier, potentially giving security teams time to respond.
AI enables operators to focus on the right cameras at the right time. A single operator of an AI-enabled surveillance system can efficiently monitor 200–500 cameras at once because the software filters out routine activity and displays only the most relevant feeds—what needs attention now, not everything being recorded.
For museum security directors, this means early intervention rather than after-the-fact evidence, fewer false positives, more efficient staffing, and lower operational costs. Museums can enhance security for high-value exhibits with custom rules for barrier breaches, or lingering visitors. And they can digitally log unusual behavior for insurance and compliance purposes.
Use Cases for AI Surveillance Beyond Art Museums
These capabilities apply to multiple sectors where small lapses lead to large losses, from museums and galleries to shopping malls and schools, from correctional facilities to critical infrastructure.
Mall security, for example, isn’t just about deterring crime: it’s about protecting reputation, retaining tenants and enabling shoppers to trust a safe shopping experience. Today’s property managers and security professionals face a troubling resurgence of retail and infrastructure crimes that threaten revenues and public perception.
And most shopping malls are forced to confront this threat with limited resources and budgets.
Now, AI-enabled surveillance systems can cover millions of square feet of stores, restaurants and entertainment venues.
Before implementing AI, control rooms primarily functioned as a review-and-report centers. Operators would replay footage after an incident to understand what happened. Now, mall security control rooms serve as real-time response hubs, enabling operators to respond instantly when threats emerge.
By automating the detection process, malls can reduce the need for large teams of operators. Instead of manually monitoring hundreds of cameras, operators can focus on resolving incidents, improving efficiency and cost-effectiveness. With AI, fewer personnel are required to support control rooms and monitor cameras more effectively.
Benefits for Security Integrators
For security integrators, dealers and managed service providers, AI-enabled surveillance systems offer a source of recurring revenue with minimal maintenance. Many software solutions are subscription-based, compatible, and customizable for a variety of cameras.
As the solution learns, it becomes easier to configure to meet the specific needs of the organization and its security teams. Most solutions will evolve with client needs and threat landscapes.
The Louvre heist was a reminder: What matters is not just what cameras can see. It’s about what security teams can understand and act upon in real time. AI-enabled surveillance won’t stop every incident. But it can deliver what traditional systems cannot: situational awareness in real time, when seconds matter most.
Kevin Brown is chief executive officer of icetana AI.
