Across the globe, the rapid pace of technology development has made it difficult to govern emerging tech effectively. Policymakers struggle with several primary issues, including knowledge of the subject matter, the potential impact on the pace of innovation, and the rapid rate of adoption. The United Kingdom’s “Secure by Design” program intends to meet these challenges, as well as take steps to position the country as “best place in the world to do digital business.” As Brexit continues, and Britain’s finance sector looks to jump ship, such a goal is as timely as it is necessary. At its core, the program will create powerful tools for policymakers, industry, consumers, retailers, and others. The final U.K. “Code of Practice” for internet-of-things security released on Oct. 14, 2018 by the Department for Digital, Culture, Media and Sport in conjunction with GCHQ’s National Cyber Security Centre offers one of the clearest policy positions articulated yet by any national government. It sets out a technically literate policy that will drive manufacturers to innovate more efficient ways to protect internet-connected consumer devices, through market and regulatory incentives.
By its own terms, the code of practice—and, more broadly, the Secure by Design program—seeks to “support all parties involved in the development, manufacturing and retail of consumer [internet-of-things devices].” To support this goal, the release is accompanied by awareness and educational documents, technical standards guidance, and an implementation plan, all of which show the U.K.’s commitment to a leadership role in securing the internet of things. The fact that the code is translated into eight languages, including Mandarin, Korean, French, German and Japanese, is crucial in showing that the U.K. intends to be a global trendsetter, but it also reflects the global nature of the markets, supply chains and security threats, as well as resilience and confidence in consumer internet-of-things devices. A common or coordinated international approach increases adoption speed, reduces transactional friction, and increases consumer confidence across global markets.
Finally, the implementation plan for the Secure by Design program demonstrates GCHQ and DCMS know well the current cybersecurity climate writ large. Like the United States, the U.K. has identified a significant shortage of trained cyber security professionals. This, compounded by the rapid development of internet-of-things devices, rollout of 5G, and other technical advances, means there is a lack of capacity to protect internet-of-things products and services from increasingly complex cybersecurity dangers. The U.K. wants to see the code ensure that devices flooding into homes and companies are equipped with necessary capabilities for owners to protect themselves—through voluntary, market-driven measures ideally, though if that fails they will “make these guidelines compulsory through law.” Consumer awareness, education, and labelling will empower buyers to make well-informed decisions and give citizens knowledge to take advantage of these capabilities. Finally, NCSC-sponsored CyberFirst summer courses will train the next generation of technology professionals to defend against security threats to internet-of-things devices.
What the Code of Practice Is—And What it Isn’t
Many early criticisms of the code are premised on a superficial understanding of the program and emerging solutions. At its core, the code details several positive, practical steps for device manufactures across the supply chain and product lifecycle. For instance, botnets like Mirai and others gain their destructive power by taking over large numbers of internet-connected computers or devices. Steps that the code recommends, like prohibiting default passwords and keeping software up to date, limit the speed and scale of a botnet’s growth, thus diminishing their ability to do harm. The code guides manufacturers away from common patterns of security failure that create openings for many types of threats, including botnets, and towards those that tend to be more successful. We outline the top three recommendations, as well as a few others worthy of note. (The paragraph numbers used below are from the document itself and do not necessarily reflect our ranking).
1. No default passwords. Passwords are meant to restrict access to systems only to those who know them. Default passwords, like “admin” or “password,” ensure anyone can know them, thus defeating their utility when defaults are published, well-known, or easily guessable. Many devices already ship with unique passwords, requiring a change on first use. Default and common passwords on internet-facing interfaces (such as Telnet and SSH) allow network worms like Mirai to propagate very quickly, though improving security of other interfaces also improves “security by default” for internet-of-things systems.
2. Implement a vulnerability disclosure policy. As more of the “things” around us depend on software and become exposed to the internet, more adversaries will take advantage of their flaws. A coordinated vulnerability disclosure policy invites allies, acting in good faith, to report these flaws to the manufacturer so they can be fixed. The device maker has an obligation to acknowledge and address issues in a timely manner. (The leading international standard for coordinated vulnerability disclosure, ISO 29147, calls for acknowledgement in 7 days. And the Code’s additional explanatory notes section makes a case for manufacturers to address the issue in less than 90 days.)
3. Keep software updated. Software updates can address bugs and vulnerabilities once manufacturers know about them. Most internet-of-things devices have the capability for software updates today, though laggards still exist. For instance, a major U.K. retailer removed GPS watches made for children over concerns that they were insecure and could not be updated. The code goes further and asks manufacturers to preserve basic functions during an update and that the update process be secured. The code also asks manufacturers to disclose a minimum timeline for software updates and makes provisions for devices or components that cannot be updated through software, noting that the manufacturer can replace them—in fact, under U.K. law they must repair or replace faulty products for 6 years.
6. Minimize exposed attack surfaces. The code aims to eliminate exposure and attack surface where the value to the consumer is outweighed by the risk associated with the vulnerability. Many devices already minimize feature sets due to resource constraints. With better hardware capabilities at lower cost, the trade offs for increasing the number of exposed services to the internet or taking a default-enable approach to elective services.
9. Make systems resilient to outages. Boosting resilience to outages will be increasingly important in coming years, as internet-of-things infrastructure changes, such as migrating to a new domain, end of life, going out of business, and other circumstances impossible to predict. Mature design processes include failure-mode analysis to guide how the device will perform during different environmental or system failures. This can mean the device notifies the person it’s in some kind of degraded mode or that mechanical systems replace software-driven ones. My Amazon Echo tells me when my internet connection (or its servers) are out, and internet-connected locks usually have mechanical keys as backups when connectivity or power are unavailable.
10. Monitor system telemetry data. Mobile phones and apps send masses of telemetry information back to their developers, who can analyze and improve the products as well as look for security or safety anomalies. Microsoft, Apple and Google detect emerging threats against their products and issue updates to address them, while doing so in a privacy-neutral way.
Threats, vulnerabilities and industry practices change over time. The code of practice is a snapshot in time, meant to be goal- or outcome-based rather than prescriptive, so organizations can adapt as necessary while still hitting these objectives. It isn’t meant to supplant technical standards, but instead the U.K. government has mapped the code to technical standards for ease of implementation.
Securing systems may increase cost of doing business, and eventually the price of consumer goods. On the other hand, buyers and owners already bear costs for insecurity. Widescale harm from events like WannaCry and NotPetya greatly escalate those costs, including harm to third-parties through no fault of their own. In conversation, U.S. and U.K. retailers have mentioned costs associated with employees educating consumers and increased rates of return for security issues.
Manufacturers are in the best position to reduce systemic cost and risk, as their available options are much greater than those of owners. Shifting responsibility and costs across the supply chain has been difficult in the past, without strong financial or regulatory incentives. The U.K. intends to drive these changes through labeling, consumer awareness, (if it must) regulation, and (I strongly suspect) by requiring devices they buy to adhere to the code.
Crucially, the Department of Digital, Culture, Media and Sport (DCMS) has made it clear that they do not intend to reinvent the wheel. An accompanying document maps the code against over 100 documents from nearly 50 organizations, representing “published standards, recommendations and guidance on [internet-of-things] security and privacy from around the world.” This is, first and foremost, an effort grounded in a practical understanding of the problem, the effective approaches, and what has failed in the past. DCMS know that manufacturers “are already implementing a range of standards,” and the mapping document shows how those efforts fit with the code.
While most of the elements of the code are well understood, some of the objectives it lays out are only beginning to enter common practice. Though they have long been technically possible, these innovative approaches have only recently been needed due to increases in attacks against internet-of-things devices and buyer pressure for higher security, largely among retailers and corporate buyers. Publishing the code will serve to stimulate innovation toward better capabilities in the hands of more manufacturers, buyers, and owners.
Most device makers should have no problem meeting the objectives laid out in the Code within the next couple of years. Several of the world’s largest manufacturers have already committed to a similar high-level doctrine called the “Charter of Trust.” The manufacturers most likely to be impacted are those that buy very low cost, low quality devices from China or elsewhere and repackage them under a variety of names. These brands tend to exit the market after only a year or two, replaced by other brands selling nearly identical-looking products from the same factories, making it hard to enforce accountability for support. They live on, connected to the internet, vulnerable and exposed to global accidents and adversaries. This kind of market confusion drives out better products leaving buyers with low choice and low quality—a market for lemons.
Yet the code is not a cure-all for every internet-of-things security concern. First, it only applies to home, or consumer-grade internet-of-things devices. Yet similar technologies—and their associated security risks—have been adopted across automotive, aviation, maritime, energy, and other sectors. Each of these has distinct ecosystems, challenges, and leverage points to evaluate if the U.K. wants to apply the code in those industries. Second, the unimplemented policies cannot make change, and the Secure by Design program is light on how the government plans to achieve market adoption. This will inevitably take resources, focus, and time that must be allocated amid a turbulent national and global political landscape. Third, global supply chains and markets demand international cooperation and collaboration. While policymakers have shied away from corralling rapidly advancing technologies, such as internet-of-things devices and 5G, they seem more willing now than ever before in key regions like North America, Europe, and China.
The code is a positive step forward for consumer IoT security and has positive traction. HP and Centrica have already formally signed on to the code, and others are likely to follow, given the resources the U.K. government seems to be putting behind adoption and enforcement. DCMS contend that at least eight of the code’s guidelines are already legally enforceable through the U.K. Data Protection Act and GDPR. Germany and the EU have begun adopting compatible (though much less effective) policies, and in the United States, California’s internet-of-things bill (SB-327) requires manufacturers to equip internet connected devices with “reasonable” and “appropriate” security features. (In the deliberations captured in the bill’s history, the legislature emphasized that security must be both reasonable and appropriate to the device, and that it’s up to the device makers to determine that. The code could serve as a good model for meeting this standard of care for consumer internet-of-things devices.) Globally, policymakers are reaching for clear guidelines and implementable solutions, coupled with adverse market pressure for companies that come up short.
Most of the internet-of-things devices that ever exist will be designed in the future. Policies like the U.K. Code of Practice are meant to be forward-looking, driving innovators toward better products. Many of the objectives it lays out are commonplace among moderate- and high-quality devices, even those at low price points. But the Code will raise the bar for all manufacturers and reduce susceptibility to cyber security, safety, and privacy issues. And it will give retailers and consumers a common measuring stick for comparing devices.