Columbia Science and Technology Law Review » liability

A System of Autonomous Cars

Brandon Ge — Tue, 09 Nov 2010 16:11:06 +0000

The Vehicle

According to a Bundle.com study, Americans spend an average of 72 minutes a day behind the wheel. Imagine the increased productivity and general welfare if that time could be spent on other tasks, such as doing work, checking email, making calls, or even sleeping. Having to spend less time keeping one’s eyes on the road and hands on a steering wheel will free up time for life’s greater pleasures.

That day may not be far away. Google has recently test-driven seven artificial intelligence-powered cars over more than 140,000 miles on California roads, including highways and Lombard Street, which is famous for its turns and steepness. For the testing, Google had two people aboard: one technician and one person in the driver’s seat ready to take over the controls if necessary. So far, the only accident occurred when another vehicle rear-ended a Google car at a red light.

It is not difficult to picture the enormous impact that self-driving cars will have on society. When the technology fully develops, it will revolutionize the way we live in the same way that television, the Internet, and cell phones have. As Google engineers have pointed out, robots react faster than humans, have 360-degree perception, and do not suffer from problems affecting human drivers that frequently result in accidents, such as distraction, drowsiness, and intoxication. On Google’s official blog, engineer Sebastian Thrun estimates that the technology can reduce traffic-related fatalities in half, a figure that I believe to be very modest when we think about the cars’ true potential.

The Route

As Kenneth Anderson notes, the real potential of autonomous cars will only be realized if we outlaw all manual driving and every car on the road drives itself. Cars would then be able to travel safely with less distance between them, which, according to Markoff’s New York Times article, would double road capacity. If a system of artificial intelligence-powered cars makes accidents a thing of the past, we will be able to use lighter materials in car manufacturing, both cutting resource consumption in the car manufacturing process, as well as increasing fuel efficiency.

Such a scenario will not be feasible if there are even a small number of human-driven cars left on the road, or even if autonomous cars include the ability to manually take over the controls. An autonomous car will still be at the mercy of its physical limitations and the constraints of the road system, and will have difficulty avoiding accidents if it shares a congested road with reckless drivers. Another problem is that computerized cars will likely employ very passive driving systems designed to avoid accidents under all circumstances. Aggressive human drivers would be able to take advantage of passive computer drivers all day.

How feasible is a system of exclusively self-driven cars? The technology will not likely be the problem. We already have cars that can parallel park with limited input from the driver, as well as cars that brake on their own if a crash is imminent. Additionally, driving can be fairly easily broken down into various algorithms and functions, as the main variables involved are predictable: lanes of standard widths and traffic lights that change in predictable intervals. Currently, the only truly unpredictable element is the behavior of other drivers which would be much more foreseeable once computers take control.

The greatest barriers to such a system might very well be legal in nature. One such barrier is liability. In a system of robot-driven cars where people are free to do other things while their cars transport them from place to place, such as checking email or napping, who should bear accident liability? Since the benefit of such technology is to allow drivers to perform other tasks while in the car, placing accident liability on them would be counterproductive. Putting liability on the technology companies might discourage the innovation needed to fully develop the technology in the first place. On the other hand, if the technology works like it is supposed to, crashes should become a rare occurrence, reducing such a burden. Another possibility is that insurance companies will bear liability.

A second issue that comes up is whether the government should be able to outlaw something as fundamental to our society as driving cars. Americans have loved their cars for decades. Many of us know people who beam with joy after a car purchase and enjoy the freedom associated with taking a car for a spin on an open road. There will undoubtedly be drivers with squeaky-clean driving records who will claim that their driving poses no threat to everyone else. There will also be people who will distrust having a computer take over their driving, especially since the consequences of a malfunction might include serious injury and death. Some may even claim that their driving is a form of expression, and that a ban violates their First Amendment rights.

However, I think the government will ultimately prevail. Congress can look to the Commerce Clause for a Constitutional source of authority in passing legislation requiring use of self-driving cars on roads. Is there a countervailing fundamental right being infringed? Ever since The Passenger Cases, the Supreme Court has recognized a basic right to interstate travel and laws prohibiting or burdening interstate travel must meet strict scrutiny. Nevertheless, it does not appear that mandating use of self-driving cars would substantially burden this right. Airplanes, trains, and perhaps even self-driving buses will still be available as modes of transportation. Furthermore, since self-driving cars will ostensibly be safer, they can be built in such a way as to accommodate the much lower likelihood of accidents, driving car prices down. This lower likelihood of accidents would drive insurance premiums down as well. Moreover, better fuel efficiency means less spending at the pump.

The Destination

Judging from Google’s promising test runs, it appears that it is not a matter of if but when a self-driving car will become available to consumers. When that day comes, we should be ready to hurdle the legal issues that pose the greatest barrier to a system of autonomous cars. Once we implement such a system, the benefits will be nothing short of extraordinary.

Prison terms for Google executives in Italy?

Brian Harley — Fri, 27 Nov 2009 22:22:41 +0000

An Italian prosecution against Google made the headlines again this week (New York Times, Bloomberg) with the news that prosecutors in Milan are pushing for three Google executives and one former executive to be sentenced to terms of imprisonment for their failure promptly to take down an offensive video from the Italian-language Google Video service in 2006. Readers in the U.S. and elsewhere may be baffled at the idea that the facts at issue should lead to prison terms. We take a look at the facts and the law, and consider whether the prosecutions are reflective of a profoundly different legal culture, raising issues about how providers of internet services will navigate diverse legal regimes in the future.

“Most entertaining video”

The case relates to the uploading to Google Video of a mobile-phone video showing an autistic high school student being bullied by his classmates. The video was allegedly uploaded to the Italian-language Google Video on September 8, 2006, and not removed until November 7. According to the prosecutors, the video appeared in the site’s “most entertaining videos” section, ranked as the twenty-ninth most viewed. The video was only taken down after Down-syndrome advocacy group Vivi Down appealed to the Italian authorities, who in turn demanded that Google take down the video (it would appear that though the victim was in fact autistic and did not have Down Syndrome, Vivi Down’s involvement was prompted by derogatory references to Down Syndrome in the offending video). According to Google, it did everything that was required of it under the applicable laws, and removed the video within hours of being notified. The four Turin youths involved in the bullying were subsequently tracked down (with the help of Google), and sentenced to one year community service with a center for children with Down syndrome.

Google: ISP, content provider, or something else?

Italian law is in line with European standards in relation to internet service provider liability: Italian legislative decree of April 9, 2003, n.70, faithfully implements EU Directive 2000/31/EC on Electronic Commerce (a European directive is a legislative instrument adopted at the EU level, which is not directly applicable in the member states, but must be separately implemented by each government). Art. 14 of the Directive and Art. 16 of Legislative Decree stipulate that an “information society service” provider who provides hosting services (defined as “storage of information provided by a recipient of the service”) shall not be liable for information stored at the request of the recipient, on condition that it does not have actual knowledge of illegal activity and that, upon obtaining such knowledge, the provider acts expeditiously to remove or disable access to the information. Art. 15 of the Directive and Art. 17 of the Italian Legislative Decree provide that there is no general obligation to monitor the information which providers transmit or store, nor a general obligation actively to seek facts or circumstances indicating illegal activity. Criminal prosecutions of internet company executives are rare, but not unprecedented in Italy (see Eric J. Lynam’s article in Privacy & Security Law).

Discussions of the case have focused on the issue of whether Google is an “internet service provider” (ISP) on the facts. If Google’s role in the provision of the Google Video service is that of an ISP, so the reasoning goes, it should be able to avail of the defenses under the Electronic Commerce Directive described above. On the other hand, if it is deemed a content provider (such as an online newspaper), it will be held responsible for the content that it hosts. In fact, both the European and Italian laws are applicable to “the provision of an information society service” (“prestazione di un servizio della società dell’informazione“), which is defined as “any service normally provided for remuneration, at a distance, by means of electronic equipment for the processing (including digital compression) and storage of data, and at the individual request of a recipient of a service.” (Article 1(2) of Directive 98/34/EC as amended by Article 1(2)(a) of Directive 98/48/EC). According to Google’s lead attorney in the case “Google Video is not a content site and it is not an ISP, it is something else entirely. … Google is an instrument people use to locate content produced by someone else. It is a mistake to try to make it fit into the definition for something different.” (quoted here). The legal question appears to be whether, in light of the nature of the service, Google could establish that (a) it is providing an “information society service”; and (b) that the service provided by Google Video amounts to “hosting.” If so, the issues should be confined to if Google had “actual knowledge” of the video.

The issue of knowledge appears to be precisely what the Milanese prosecutors are focusing on: they argue that Google must have known of the existence of the video well before November 7. According to their submissions, in light of comments posted to Google Video voicing outrage at the video, “[i]t is reasonable to imagine that comments like this were followed by requests by these same people that the video be removed” (quoted in the New York Times article). If it were indeed the case that Google had actual knowledge of the video (which they vigorously deny), and delayed in taking it down, then a finding of liability would not be particularly inconsistent with web hosting liability laws on either side of the Atlantic.

Calls for filtering?

There are, however, suggestions that the Milanese prosecutors are calling for more than prompt take-downs of offensive material. According to La Stampa, they are arguing that Google could relatively easily implement “controls” to prevent such incidents occuring again in the future. Rather disturbingly, they point to the example of Google’s deal with the Chinese authorities to provide a “censored search engine for Chinese use.” Are the Milanese prosecutors really openly calling for the kind of active filtering imposed by the Chinese authorities? Also rather disquieting is the prosecutors’ conclusion that “Google only implements filters when it sees an opportunity for gain,” and its comment that Google’s choice of a freely accessible service was motivated by its desire to increase its revenues by maximizing the diffusion of videos hosted on its services. Is the profit-motive on trial here? The prosecution seems to be suggesting that Google could easily have implemented an effective filter, but refrained from doing so out of an a concern that it might impact on its profits. But on what grounds would Google have seen itself as obligated to implement such a filter, where prompt take-down policies have to date generally been deemed adequately to protect those harmed by offensive content, in Europe and elsewhere?

Vivi Down, the advocacy group which was instrumental in bringing the video to the attention of the Italian authorities, has stated in a press release:

“Vivi Down has no desire to see the Internet censored, but acts out of respect for the legitimate rights of a party harmed by a criminal offense, so that the judicial authorities can definitely ascertain whether the publication of the video involved the commission of one or more crimes precisely provided for by our criminal legislation. Within a democratic society, freedom of expression is sacrosanct, just as is the respect of the rules upon which social cohabitation is founded and respect of the rights of one’s neighbor, especially of those of the weak and defenseless.”

Clearly, there is a balance to be struck between freedom of expression (both online and off-line) and the rights of others, but the approach the Milanese prosecutors seem to be advocating, characterizing Google as a content provider with direct responsibility for all content posted, implies striking that balance in favor of extensive obligations to monitor and filter online content, which would undoubtedly have a chilling effect on online freedom of expression.

According to the New York Times, even if found guilty, the four defendants would not in fact serve prison sentences, as prison sentences of less than three years are commuted in the absence of a criminal record. Yet the allegations are serious in nature, and a conviction would clearly compel Google to reassess how it delivers its services in Italy and elsewhere, possibly even forcing it to implement some kind of active filtering of content, and making the task of providing online services across different jurisdictions ever more challenging.

By Brian Harley.