• United States



Rise of the AI Overlord: Machines monitor, automatically detect suspicious behavior

Aug 21, 20126 mins
Data and Information SecurityEnterprise ApplicationsMicrosoft

While TrapWire seems to use 'behavioral recognition' to analyze video and camera feeds, a 'video camera on its own is dumb.' BRS Labs has software with an extraordinary AI brain that can see, learn, get smarter with time, and make decisions on what behavior recognition threats to report in real-time. It needs no human judgement to spot bad guys; the military-grade technology can automatically detect and alert authorities to suspicious behavior. Does that feel less like a privacy invasion?

abstract background light blue
Credit: Illus_man/Shutterstock

When it comes to monitoring CCTV video feeds for suspicious activity, the human brain reportedly overlooks 45% of all activity after 12 minutes. After 22 minutes, the human brain overlooks 95% of all activity. But what if all the 45 – 60 million eye-in-the-sky cameras were connected to computers with artificial intelligence . . . computers with an extraordinary AI brain that can see, learn, get smarter with time, and make decisions on what behavior recognition threats to report in real-time?

While TrapWire seems to use “behavioral recognition” to analyze video and camera feeds, a “video camera on its own is dumb.” Most video analytics are rule-based logic systems that continually need the rules redefined, can give hundreds of false-alarm alerts, and even miss true threats in real-time. “We are seeing more and more surveillance cameras installed everywhere, and increasingly they are being networked together. As artificial intelligence improves, video analytics may become capable of tracking increasingly complicated behavior,” the ACLU reported. “Ultimately, we need to confront the central question facing us: how are we going to handle the increasing capability of machines to monitor us in ways large and small, wide and deep?”

Globally, there are “more than 45 million CCTV surveillance systems,” according to Homeland Security Research.” This decade “will be marked by the fusion of CCTV with Biometrics, and human behavioral signatures, which will create a new multibillion premium security market of CCTV-Based Remote Biometric & Behavioral Suspect Detection.” That market is forecasted to grow from $750 million in 2011 to $3.2 billion by 2016.

As of right now, the smartest AI suspicious behavioral recognition seems to be a system with military-grade technology that “has the capability to learn from what it observes, remember activity patterns and adjust to changes in the environment, field of view and equipment – without manual interaction.” That system with AiSight and Hypocepts, which allows it to “build memories and hypothetical concepts,” is the brainchild of Houston-based BRS Labs. This year at the Counter Terror Expo in London, John Frazzini, President of BRS Labs [PDF], said, “Being recognized by the security industry with the 2012 Counter Terrorism#mce_temp_url# and Security Specialist award for video surveillance innovation underscores the game changing technology that BRS Labs is delivering to#mce_temp_url# the video surveillance marketplace.”

The system is setup at Port Fourchon, located on the Gulf Coast in Louisiana, to give first responders ‘as they happen’ alerts that are “identified automatically by the surveillance system” so they “can coordinate their response in real time.” It’s also in Houston which, according to Hobby Wright, Vice President for Strategic Programs for BRS Labs, “is one of many American cities deploying and developing programs that incorporate our intelligence into their video surveillance operations.” El Paso established a Security Alert Monitoring (SAM) center that will use the BRS Labs system [PDF] to monitor water treatment plants adjacent to the Mexican border. Upon completion, the SAM Center “system will be made available via the Internet to the El Paso Fusion Center,” law enforcement and federal authorities like the US Border Patrol. It’s also set to guard the World Trade Center and “will be connected to the NYPD’s Lower Manhattan Security Initiative.” The system is expected to be completed in 2013 and will reportedly “cost tens of millions of dollars.” But Frazzini told The Post, it “is light years ahead of the old-fashioned security cameras monitored by night watchmen everywhere.”

Tampa wanted a video management system for the 2012 Republican National Convention which would “be able to track at least 300 moving objects within a single frame, monitor video feeds from at least 25 cameras simultaneously and give remote access to up to 150 users.” At that time, John Dingfelder, the ACLU’s senior staff attorney for mid Florida was concerned about using the cameras after the convention. He said, “I don’t think that that’s the kind of community that we want to be, under constant surveillance, especially constant surveillance by the government.” Even if BRS Labs didn’t win the contract for Tampa, a $2 million deal will see the system deployed to look for bad guys in California.

According to the San Francisco Municipal Transit Authority [PDF], the BRS Labs system will be at 12 MTA train stations and is capable of “tracking over 150 objects and activities on a continuous basis.” No wonder it won the contract since it has capabilities far beyond tracking 150 objects at a time. After BRS Labs won the Government Security News Award for Best Intelligent Video Surveillance Solution, we looked at an AI controlled video surveillance society, a mix between real life HAL 9000 meets Skynet. Frazzini had said that video analytics is “dead” and fatally flawed, and “open-sourced algorithms” lack “intelligence to understand what it is seeing.” He added, “The AISight 3.0 solution can handle over 500 video feeds and can detect 350 objects per camera field of vision. The system multiplies a surveillance systems ability to detect anomalies consistently.”

Earlier this year, BRS Labs was granted a patent for its AISight 3.0 video surveillance software platform that enables “a video surveillance system to recognize complex behaviors by analyzing pixel data using alternating layers of clustering and sequencing.”

In addition to the behavioral recognition system patent, other BRS Labs’ intellectual property filings cover technical breakthroughs in background models, detection, tracking, object characterization, classification, scene characterization, target matching, techniques for unsupervised learning of spatial and temporal behavior, long term associative memories, anomaly detection using long-term memories, sudden illumination change, scene preset identification, trajectory learning, trajectory anomaly detection, spatial and temporal anomaly detection, clustering techniques in self organizing maps, classification anomalies, semantic representation of scene content, and a cognitive model for behavior recognition.

Check out the video below to learn more about BRS Labs analytics to detect suspicious behavior.

It’s the rise of the smart machines to automatically detect and report suspicious behavior. At Def Con’s Bigger Monster, Weaker Chains, both NSA whistleblower William Binney and James Bamford agreed that the NSA is playing word games when it comes to domestic spying, but “it's ‘technically legal’ so ‘long as no human listens to or reads any of the harvested communications without a warrant’.” Do you feel better about privacy if it’s a machine doing the monitoring and not human eyes? It’s not going away; AI machines automatically monitoring CCTV feeds for suspicious behavior is projected to grow into a $3.2 billion industry by 2016. Do you welcome your AI Overlord?

ms smith

Ms. Smith (not her real name) is a freelance writer and programmer with a special and somewhat personal interest in IT privacy and security issues. She focuses on the unique challenges of maintaining privacy and security, both for individuals and enterprises. She has worked as a journalist and has also penned many technical papers and guides covering various technologies. Smith is herself a self-described privacy and security freak.