Internet of Things security check: How 3 smart devices can be dumb about the risks

Internet of Things security is no longer a foggy future issue, as more and more such devices enter the market--and our lives. From self-parking cars to home automation systems to wearable smart devices, analysts currently estimate that some 50 billion to 200 billion devices could be connected to the Internet in 2020. Google CEO Eric Schmidt told world leaders at the World Economic Forum in Davos, Switzerland, in January, "there will be so many sensors, so many devices, that you won't even sense it, it will be all around you," he said. "It will be part of your presence all the time."

That's hardly comforting when you consider how many of these smart devices still seem to be pretty dumb about security. A study by HP of ten popular IoT devices--including smart TVs, webcams and home automation devices--found an average of 25 security flaws per device. Seven of the ten devices had serious vulnerabilities.

Three smart devices with dumb security risks

Three of the hottest IoT categories offer examples of the risks. The Withings Activité Pop, shown off at CES, is an analog watch that records a user's daily habits, including sleep time, steps, swimming and other activities. Yet one Symantec study of sports bands and smart watches found the majority lacked privacy policies, nearly all connected to a cloud service, and 20 percent sent passwords without encrypting them.

Home automation faces the same issues. A 2013 report, for example, found significant vulnerabilities and poorly secured default configurations in home automation from vendors such as Belkin, Insteon, Linksys and Sonos. One intrepid reporter even used the information from the research to contact users and demonstrate that she could control their homes.

Some manufacturers are listening. Lock maker Kwikset, for example, has created a touchscreen deadbolt lock that uses a smudge-resistant touchpad, making it more difficult for an attacker to attempt to discern a homeowner's code from fingerprint residue left on the keypad.

Often, however, policymakers have to prod the industry to create better protections for consumers. Take BMW's self-parking car demoed at CES (and expected to come out as a feature about 2020). A car that can drive itself could also be hijacked by attackers. In a presentation at the Black Hat security conference in Las Vegas last August, two researchers--Charlie Miller, a security engineer with Twitter, and Chris Valasek, director of vehicle security research with security services firm ioActive--studied 19 different models of cars and found vulnerabilities in every vehicle. The researchers also showed that they could take control of the cars with self-steering mechanisms.

Even Congress has weighed in on this particular danger. On February 9, Senator Edward J. Markey, D-Mass., released a report claiming the automotive industry was not even close to securing its vehicles from cyberattacks. Almost all cars currently being sold have wireless technologies, the foremost application of which is to connect tire-pressure monitoring devices to the brains of the car. Yet, few manufacturers had taken steps to prevent remote access, the report concluded.

Vendors pay spotty attention to IoT security

Candid Wueest, principal security engineer with Symantec and the author of the Symantec report, stressed that often there is little consumers can do about IoT security, except to urge vendors to take it more seriously.

"Vendors are saying that they could implement security, but no one is asking for it," he says. "So, if no one is going to pay for it, it is not on their list of priorities."

David Grier, an associate professor of International Science and Technology Policy at George Washington University and past president of the Institute of Electrical and Electronics Engineers (IEEE), echoed Wueest's expectation that the industry would be reactive rather than proactive. "Everything in security in the past has required an incident," he says. "That gets people focused on the issue. At some point, you have to not only demonstrate the nature of the problem, but how it hurts."