Key Takeaways
- Small‑business owners are turning to AI‑powered video analytics to combat shoplifting, citing mounting losses and the fatigue of constant manual monitoring.
- Veesion’s software plugs into existing security cameras, detects suspicious gestures (e.g., concealing items), and alerts staff within roughly 25 seconds.
- Early adopters report quicker interventions and the ability to ban repeat offenders, though concerns linger about false accusations and the technology’s reliability.
- Critics argue that AI‑driven theft detection may disproportionately flag innocent shoppers based on benign movements, calling for greater transparency and oversight.
- Proponents maintain that the AI does not collect biometric data and acts only as an assistive tool, leaving final decisions to human employees.
The Growing Pain of Shoplifting for Independent Retailers
Manny Singh, manager of a BP gas station on S Keystone Ave in Indianapolis, described the relentless strain of watching a tiny surveillance monitor for signs of theft. “Every five seconds I have to look at the (screen) to see what everybody’s doing,” he said, noting that the constant vigilance is both exhausting and ineffective. Despite a national dip in retail theft—Capital One Shopping Research reported a 10 percent drop in 2025—small businesses continue to feel the pinch. Singh estimated that shoplifting costs his station “thousands, every month,” a figure that became intolerable after a brazen incident: “That’s like $90 worth of candy gone in like two seconds. That’s when I realized I need to do something.” His experience mirrors that of many owners who find traditional loss‑prevention methods insufficient against opportunistic thieves.
How Veesion’s AI Integrates with Existing Cameras
To address the problem, Singh’s station adopted Veesion, an artificial‑intelligence platform designed to augment ordinary security feeds. The software overlays directly onto current camera systems, analyzing footage in real time for behaviors commonly associated with shoplifting—such as slipping items into bags, pockets, or clothing. When the algorithm detects a suspicious gesture, it flags the event and pushes a notification to a designated employee’s phone or tablet, accompanied by a short video clip of the activity. Veesion claims the average lag between the flagged action and the alert is about 25 seconds, a window that allows staff to intervene before the thief exits the store. The technology does not replace human judgment; rather, it supplies an extra set of eyes that works continuously without fatigue.
Real‑World Test Shows Prompt Alerts
The station’s I‑Team conducted a live test to gauge Veesion’s responsiveness. By concealing a bag of chips inside a backpack and walking past the cameras, the team triggered an alert that arrived on Singh’s phone “almost instantaneously.” Singh recounted the outcome: “I’ve found success using Veesion to confront people taking items, or banning them from the store if they’re recorded stealing.” The immediate feedback enabled him to approach the suspect, recover the merchandise, and, when appropriate, issue a store‑wide ban. For Singh, the system has turned a reactive loss‑prevention posture into a proactive one, reducing the frequency of repeat offenses and providing documented evidence that can be shared with law‑enforcement if needed.
Concerns Over Accuracy and Potential Bias
Not everyone views the technology as a panacea. Will Owen of the Surveillance Technology Oversight Project (STOP) warned that AI‑based shoplifting detection risks becoming “pseudo‑science” that unfairly targets innocent patrons. “No shopper should have to be concerned that going into a store will falsely flag them for the way that they walk, gesture, or move,” Owen argued. He cautioned that algorithms trained on limited datasets might misinterpret benign actions—such as adjusting a coat or reaching for a wallet—as theft precursors, leading to embarrassing confrontations or wrongful accusations. Owen urged businesses to adopt clear policies, provide staff training on interpreting alerts, and maintain avenues for customers to contest false positives.
Veesion’s Safeguards and the Role of Human Oversight
In response to such critiques, Hiren Mowji, Veesion’s head of sales, emphasized that the system is designed to be privacy‑conscious and decision‑neutral. “Detection is anonymous and doesn’t collect ‘biometric information’, like race or sex,” Mowji said. He clarified that the AI merely highlights anomalous movements; a human employee must view the attached video and decide whether to act. “AI can just perform more consistently and more accurately than a human can,” Mowji added. “You can sort of lower your shoulders a little bit and let the AI do the work.” This division of labor, according to Veesion, mitigates the risk of fully automated accusations while still delivering the efficiency gains that overwhelmed small‑business operators seek.
Broader Adoption and Outlook for Small‑Business Security
Singh noted that the BP station is not an isolated case; Veesion is now deployed across several locations operated by his parent company, Fateh Company. He expressed satisfaction with having “an added eye looking out for my business so I don’t have to watch profits walking out the door.” As more independent retailers grapple with thin margins and limited security budgets, AI‑assisted surveillance offers a compelling middle ground between costly guard services and ineffective manual monitoring. Nevertheless, the technology’s long‑term success will hinge on balancing its deterrent benefits with robust safeguards against misuse, ensuring that the pursuit of loss prevention does not erode trust between stores and the communities they serve.
Indianapolis businesses use artificial intelligence to stop shoplifting