Read about it here.

I wrote a post for the Harvard Journal of Law & Technology Digest outlining the California bill, but I would like to air one of my personal concerns here. As with the Nevada legislation, I am concerned that driverless cars can be used to gather personal information. The Ninth Circuit has already been struggling with the FBI’s desire to use technologies in cars to secretly gather information for law enforcement purposes. The Supreme Court also recently considered the use of GPS trackers in criminal investigations. However, cars operated entirely by computers have the potential of providing much more sensitive information. This information could get exposed not only to government surveillance, but also to data mining companies or hackers that could tap into the driverless systems to take control over the cars. The California bill should therefore explicitly require the Department of Highway Patrol to adopt standards with respect to privacy and security, in addition to “safety standards and performance requirements.”

Read about it here.

AB 511 defines an “autonomous vehicle” to mean “a motor vehicle that uses artificial intelligence, sensors and global positioning system coordinates to drive itself without the active intervention of a human operator.” It defines “artificial intelligence” to mean “the use of computers and related equipment to enable a machine to duplicate or mimic the behavior of human beings.” This definition, as discussed in Ryan Calo’s blog, fails to acknowledge the fact that certain vehicles already include “artificial intelligence” to various degrees. This legislation may, for example, unintentionally require special testing of self-parking cars that are already available for public consumption and are not commonly considered to be “autonomous vehicles.”

The Legislation also lacks analysis regarding collection of data from vehicles operated by computers. There is a clear advantage for a company that ultimately will provide these cars (whether it is Google or not) to collect data from the cars to advance the performance of the technology. There should be some clear directions as to how a company is to handle that data. It may, for example, be kept in a manner that does not reflect personal information about where any particular vehicle has been or is going. More importantly, there needs to be restrictions on the government’s ability to access that data. This is significant given that the government already tried to use the current technology in certain cars to obtain information for “law enforcement purposes” in a rather non-transparent manner. Restrictions on government surveillance are usually developed by case law. But courts have been slow to develop principles to address computers and new technology. Therefore, it may be wise to address these issues with legislation based on thorough input from engineers and researchers that can provide insight into the potential of this technology.

Obviously, these issues take a long time to consider and they should not get in the way of the development and use of this technology. However, because autonomous vehicles were not previously expressly prohibited in Nevada, they could probably have been used even before this legislation (though there is some debate about whether the lack of prohibition means that autonomous cars are legal). It is, however, clear that the Nevada Legislature dropped all practical considerations into the DMV’s lap, which definitely lacks the technical expertise and resources to consider these issues (if the California DMV’s workload is any indication, early this year, DMV’s staff estimated the process of printing new driver’s licenses to average about 8 months per license!). Given that this technology will likely not be available within the next year, the Nevada Legislature could have taken some time, working with experts (perhaps Google?), to consider these issues.