Back to Index of Regulation of Artificial Intelligence

Brazil

During the 2018 discussions of the GGE on LAWS that was established in 2016 during the Fifth Review of the UN Conference of the High Contracting Parties to the Convention on Certain Conventional Weapons, Brazil issued a joint statement along with Austria and Chile, which proposed to establish an open-ended GGE to negotiate a legally binding instrument to ensure meaningful human control over critical functions in LAWS.[1]

I. Data Protection Law

On August 14, 2018, Brazil enacted Law No. 13,709, the General Data Protection Law (Lei Geral de Proteção de Dados), which aims to give more security to consumers of technology.[2] Article 1 of the Law provides for the processing of personal data, including digital media, by either a natural person or a public or private legal entity, for the purpose of protecting a person’s fundamental rights of freedom, privacy, and free development of personality.[3]

The protection of personal data is based on respect for privacy; informational self-determination; freedom of expression, information, communication, and opinion; the inviolability of intimacy, honor, and image; economic and technological development and innovation; free enterprise, free competition, and consumer protection; and human rights, free development of personality, dignity, and the exercise of citizenship by natural persons.[4]

II. Trade Negotiations

The Brazilian arm of the International Chamber of Commerce and the UN Conference on Trade and Development signed an agreement to work together to harness the power of cutting-edge technologies such as AI and blockchain to enhance and improve trade. [5] With the growing complexity of international trade agreements, the purpose of the use of AI is to reduce such complexity and help representatives of less favored nations achieve better results.[6]

III. Fraud Detection

The impossibility of inspecting all customs operations and identifying all fraud led the Brazilian Federal Revenue Service to establish partnerships with Brazilian universities to start making use of AI to detect such practices.[7] The idea is to develop a system to help customs officers identify suspicious customs operations, and to develop a product and foreign exporter information system to help importers in the registration and classification of their products and corresponding exporters.[8]

IV. Corruption Fight

The Ministry of Transparency and Controller General’s Office (Ministério da Transparência e Controladoria-Geral da União, CGU) implemented a system to find evidence of deviations in the performance of public servants. The software uses machine-learning features, an AI technique that feeds data, presents criteria, and checks if the results of the analyses performed by the machine are within the expected range. [9] Another system created by the CGU to combat irregularities is intended to oversee contracts and suppliers. The tool analyzes the risks, including not only corruption, but also other problems such as failures to comply with an agreement.[10]

V. Criminal Investigation

The Public Prosecutor’s Office of the State of Rio de Janeiro has decided to invest in advanced technologies to expedite investigations and prevent crimes from going unpunished. [11] The investment is geared towards data science and AI to collect, store, and analyze large volumes of information.[12] The system allows information from different sources and bodies to be collected and also allows a series of real-time data to be collected from suspected criminals.[13]

Canada

I. Background

Canada’s approach to AI appears to be focused on funding research as opposed to developing regulations and governance structures.[14]Canada has the “second largest tech sector outside Silicon Valley” and is considered a global leader in the field of AI.[15] In 2017, estimates indicated that funding raised by Canadian AI companies would “exceed US$250 million, representing an almost two-fold increase from the previous record historical high of US$143 million in 2015.”[16]

Deloitte’s AI practice published a recent report that raised the concern that Canada is falling behind in respect to demand and actually adopting the technology, stating, “[w]hile Canada’s AI startup environment and talent pool is relatively strong, it’s the other component of a leading AI economy—the need for robust demand—that is of greatest concern.”[17] One report by The Logic found that Canadian firms have been filing fewer AI patents each year since 2016 despite large government investment.[18]

II. Pan-Canadian AI Strategy

In the 2017 Budget, the Canadian federal government provided CA$125 million (about US$93.3 million) to launch the Pan-Canadian Artificial Intelligence Strategy.[19] This made Canada the first country to release a national strategy for AI, although the strategy is primarily focused on research and talent.[20] The strategy is being led by the Canadian Institute for Advanced Research (CIFAR), a nonprofit, private institute of advanced study that is partly government funded, in collaboration with “three newly established AI institutes”—the Alberta Machine Intelligence Institute (Amii) in Edmonton, Mila in Montreal, and the Vector Institute in Toronto.[21] The strategy is intended to further build on Canada’s AI research-based ecosystem[22] by pursuing four major goals:

• To increase the number of outstanding artificial intelligence researchers and skilled graduates in Canada.
• To establish interconnected nodes of scientific excellence in Canada’s three major centres for artificial intelligence in Edmonton, Montreal and Toronto.
• To develop global thought leadership on the economic, ethical, policy and legal implications of advances in artificial intelligence.
• To support a national research community on artificial intelligence.[23]

The CIFAR AI and Society Program “examines the policy and ethical implications of AI.”[24]

III. Regulatory Framework and Proposed Changes

A. Automated Vehicles

In Canada, motor vehicle transportation is a “shared responsibility between federal, provincial and territorial governments.” [25] Provinces and territories have jurisdiction over “approving and overseeing trials of automated vehicles that take place within their jurisdiction.”[26] Two provinces have enacted changes to allow testing for automated vehicles. On January 1, 2016, Ontario’s Ministry of Transportation launched a pilot to allow the testing of automated vehicles though the promulgation of a regulation under Ontario’s Highway Traffic Act.[27] The Ontario government“requires prior approval to test vehicles under the pilot project.”[28] One of the requirements of the Ontario’s Automated Vehicle Pilot Program is that “[t]he driver must remain in the driver's seat of the vehicle at all times and monitor the vehicle’s operation, unless approved for driverless testing.”[29] On April 18, 2018, assent was granted to Bill No. 165, amending the Québec Highway Safety Code to similarly allow a testing pilot for automated vehicles.[30] The first pilot project for autonomous buses and minibuses was authorized by an order of the Minister of Transport, Sustainable Mobility and Transport Electrification.[31] The order requires that “A driver must be able to immediately take over the driving of the vehicle should the need arise.”[32]

The federal Senate Committee on Transport and Communications released a report in January 2018, which found that “Canada is ill-prepared for the fast-approaching future of transportation.”[33] The report set out sixteen recommendations, including developing a coordinated national strategy. The strategy would “allow the government to prevent potential harms. Strong cybersecurity measures will be necessary to maintain public safety and confidence, and rigorous oversight is required to ensure personal information gleaned from connected and automated vehicles is securely held and not exploited.”[34]

B. Privacy and Transparency Challenges

There are a number of laws at the federal and provincial levels in Canada that relate to the protection of personal information. The Personal Information Protection and Electronic Documents Act (PIPEDA) is a federal privacy law that is applicable to the private sector.[35] At a February 2017 appearance before the House of Commons Standing Committee on Access to Information, Privacy and Ethics, the Privacy Commissioner of Canada stated that

[c]onsent has always been considered a foundational element of PIPEDA. Legally, organizations must obtain consent to collect, use and disclose an individual’s personal information, subject to a list of specific exceptions. But obtaining meaningful consent has become increasingly challenging in the age of big data, the Internet of Things, artificial intelligence and robotics.[36]

The Standing Committee released a report on February 28, 2018, that included recommendations to update PIPEDA.[37] The report is said to be “heavily influenced by the direction set in the European Union General Data Protection Regulation.”[38] It expressed concerns over transparency of AI decision-making (“users have little information about how they work, the data they collect and how they are used”[39]), and the risk of algorithms using personal information to “perpetuate prejudices or discriminatory practices.”[40] One of the key recommendations of the report was that “the Government of Canada consider implementing measures to improve  algorithmic transparency.”[41]

On May 24, 2018, the Office of the Privacy Commissioner of Canada published two guidance documents, one on obtaining meaningful consent and the other on inappropriate data practices, to help organizations ensure they comply with their privacy obligations in the digital age. On announcing the publications, the Commissioner stated that, “[d]uring an extensive public consultation, we heard very clearly that the increasingly complex digital environment—with technological innovations such as big data, the Internet of Things and artificial intelligence—is posing challenges for privacy protection and the consent model.”[42]

C. Directive on Automated Decision-Making for Federal Institutions

According to the government of Canada, it is “increasingly looking to utilize technology and automated systems to make, or assist in making, administrative decisions to improve service delivery.” [43] Under the authority of the Financial Administration Act, the Treasury Board of Canada issued a Directive on Automated Decision-Making, which took effect on November 26, 2018.[44] According to the Cyberjustice Laboratory of the University of Montreal,

[the Directive] outlines the responsibilities of federal institutions using AI-automated decision systems. Supporting a host of policies in the federal public administration, the Directive aims at helping to better understand and better ensure an ethical and responsible implementation of AI. Compliance with its requirements is expected from all federal institutions by no later than April 1, 2020.[45]

The Directive applies to the use of automated decision systems that “provide external services and recommendations about a particular client, or whether an application should be approved or denied.”[46] The Directive provides for a questionnaire, called an Algorithmic Impact Assessment, “designed to help [federal institutions] assess and mitigate the risks associated with deploying an automated decision system,” [47] which should be “completed prior to the production of any Automated Decision System to be used in federal administration.”[48]

IV. International Cooperation

A. G7 Discussions

A 2018 report by Innovation, Science and Economic Development Canada states that, in the context of 2018 G7 Presidency,

[i]n March 2018, Innovation Ministers expressed a shared vision of human-centric artificial intelligence for innovation and economic growth, and released a statement that confirms the need to safeguard privacy. This vision was also expressed by Leaders in their Common Vision Statement on Artificial Intelligence, and the G7 Leaders’ Statement released in June 2018.[49]

On June 7, 2018, the governments of Canada and France released a joint statement on AI calling for “the creation of an international study group that can become a global point of reference for understanding and sharing research results on artificial intelligence issues and best practices.”[50] The statement stated that Canada and France “wish to promote a vision of human-centric artificial intelligence grounded in human rights, inclusion, diversity, innovation and economic growth.”[51] The two countries established a task force “in order to make recommendations on the scope, governance and implementation of the international study group. By the end of the year, the task force will submit a report on the implementation of the international study group, whose results will be shared within the G7.”[52]

On October 17, 2018, the Minister of Innovation, Science and Economic Development announced that he would host a G7 Multi-stakeholder Conference on Artificial Intelligence on December 6, 2018.[53] During the conference, Prime Minister Justin Trudeau further discussed the International Panel on Artificial Intelligence and also announced an “investment of up to nearly [Can]$230 million for the AI-Powered Supply Chains Supercluster (SCALE.AI) based in Quebec.”[54]

B. Lethal Autonomous Weapons Systems

At the 2017 meeting of the GGE on LAWS, Canada stated that it was “committed to maintaining appropriate human involvement in use of military capabilities that can exert lethal force.”[55] At the 2018 GGE meeting Canada was “supportive of developing key Transparency and Confidence- Building Measures and looks forward to exploring these and other such ideas here.”[56]

Jamaica

The Jamaican Government has reportedly been “looking to position Jamaica to take advantage of the benefits of artificial intelligence (AI) in boosting economic growth and job creation.” [57] According to Science, Energy and Technology Minister Andrew Wheatley, there have been ongoing discussion involving the Ministry, the Jamaica Information Technology and Services Alliance, the Jamaica Computer Society, and the Business Process Industry Association of Jamaica on how AI can best be utilized to benefit the country. In Minister Wheatley’s opinion, “while AI poses a potential threat to traditional call center operations in the business process outsourcing (BPO) sector, it presents opportunities in the high-skill areas of programming and development.”[58] He therefore called for increasing the number of computer science graduates from Jamaican educational institutions, so that Jamaica “can take advantage of the clear opportunities that will come from the new demand for highly skilled programmers.”[59]

No specific regulations related to aspects of AI technology applications were located.

Mexico

I. National AI Strategy and Directorate

In March 2018, the Mexican government announced the adoption of a strategy on AI. [60] The strategy provided for the creation of a Directorate on Artificial Intelligence tasked with the development of a framework aimed at fostering a multi-sector dialogue on AI.[61]

The Directorate was created in April 2018, within Mexico’s Commission for the Development of Electronic Government, which “seeks to support Federal Public Administration agencies in the development, adoption and use of Artificial Intelligence for the design, implementation and evaluation of public programs, policies and services.”[62]

In November 2018, the Mexican government published a document that includes general principles for the development and use of systems based on AI in Mexico’s federal government.[63] Some of the most salient principles are as follows:

• Monitor and evaluate the impacts of AI systems in order to ensure that they achieve the expected results
• Promote transparency, by explaining to the users that interact with AI systems the decision process taken by such systems, the expected benefits as well as the potential risks derived from using such systems
• Protect privacy, by incorporating mechanisms of control and consent for the use of personal data during the design of AI systems
• Foster equality, by reducing risks of discriminatory biases derived from the utilized data
• Due process, by allowing individuals to dispute decisions made by AI systems[64]

The presidential administration that adopted this strategy and its guiding principles ended on November 30, 2018. No information could be located on whether the new administration (which commenced on December 1, 2018) will continue with this strategy and its principles, or initiate a similar effort.

II. Lethal Autonomous Weapons Systems

According to information provided by the United Nations Office for Disarmament Affairs, the Mexican government has expressed its opposition to LAWS, as follows:

On 16 October 2017, the Permanent Mission to the United Nations of Mexico partnered with the International Committee for Robot Arms Control, Human Rights Watch, Seguridad Humana en Latinoamérica y el Caribe and the Campaign to Stop Killer Robots to host a panel discussion entitled “Pathways to Banning Fully Autonomous Weapons” as part of the First Committee side event series for the 72nd Session General Assembly. Ambassador Juan Sandoval Mendiolea, Deputy Permanent Representative of Mexico to the United Nations, introduced the discussion by noting that Mexico has joined other states in calling for a ban on lethal autonomous weapons systems, also known as fully autonomous weapons. . . . Ambassador Mendiolea stressed the importance of continued research and education on these weapons systems and their possible dangers.[65]

III. Autonomous Vehicles

A recent study on autonomous vehicles (AV) readiness states that, as of 2018, Mexico has a number of obstacles to adopting this technology:

AV adoption in Mexico currently faces a range of barriers, with a lack of specific regulations, no active tests and little industrial activity. . . . On policy and legislation, it scores low on both AV-specific and general variables, with no apparent regulations on AV testing on public roads and very limited testing so far. The country’s current economic and political condition creates barriers for taking actions in order to adopt AV in the near future. It gets the lowest scores on industry partnerships, research and development AV hubs, patents and investments and the research has found no AV technology company headquarters . . . . On infrastructure, Mexico’s roads receive a slightly below average rating for road quality, but lower scores on other variables. There are no active AV tests, contributing to its low rating on consumer acceptance.[66]

United States

United States lawmakers and regulators have mainly pursued AI in the area of autonomous or self- driving vehicles. The Department of Transportation is investigating what elements must be considered in drafting regulations for the use of such vehicles, including multi-vehicle convoys, and several states have adopted legislation and regulations allowing for the testing of autonomous vehicles. In addition, recent federal legislation has tasked part of the Department of Defense with the responsibility of crafting policies for the development and deployment of AI systems as they concern national defense.

I. Federal Legislation and Regulatory Action

In the 115th Congress, thirty-nine bills have been introduced that have the phrase “artificial intelligence” in the text of the bill.[67] Four of these bills have been enacted into law. Section 238 of the John S. McCain National Defense Authorization Act for Fiscal Year 2019 directs the Department of Defense to undertake several activities regarding AI.[68] Subsection (b) requires the Secretary of Defense to appoint a coordinator who will oversee and direct the activities of the Department “relating to the development and demonstration of artificial intelligence and machine learning.” Subsection (g) provides the following definition of AI:

(g) ARTIFICIAL INTELLIGENCE DEFINED—In this section, the term “artificial intelligence” includes the following:

(1) Any artificial system that performs tasks under varying and unpredictable circumstance without significant human oversight, or that can learn from experience and improve performance when exposed to data sets.

(2) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception,  cognition,  planning,  learning,   communication,   or physical action.

(3) An artificial system designed to think or act like a human, Including cognitive architectures and neural networks.

(4) A set of techniques, including machine learning, that is designed to approximate a cognitive task.

(5) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decision making, and acting.

Subsection (f) instructs the Secretary of Defense to “delineate a definition of the term ‘artificial intelligence’ for use within the Department” no later than one year after the law’s enactment.

In 2018, Congress also advised the Federal Aviation Administration, in language inserted into the FAA Reauthorization Act of 2018, to periodically review, with industry and academic experts, the state of AI in aviation and take steps, as needed, to address new developments.[69]

In 2015, Congress included language in the Fixing America’s Surface Transportation Act (FAST ACT) directing the Department of Transportation to fund research on the application of autonomous vehicles in developing and improving traffic patterns on highways.[70]

The United States Department of Transportation has solicited requests for comments for proposed studies of automated driving systems. One such request, issued by the Federal Highway Administration, asked for comments on “a range of issues related to assessing the infrastructure requirements and standards that may be necessary for enabling safe and efficient operations of ADS [Automated Driving Systems].”[71]

Another request for comment, issued by the National Highway Traffic Safety Administration, concerned possible barriers created by existing Federal Motor Vehicle Safety Standards to the testing of autonomous vehicles, especially those with nontraditional interior design elements such as vehicles that do not have steering wheels.[72]

II. State Legislation and Regulatory Action

In 2011, Nevada adopted the first legislation concerning the testing of autonomous vehicles.[73] The specific law defines autonomous vehicle as restricted to the operation of “the motor vehicle without active control or monitoring of a human operator.”[74] The law also sets forth requirements for the testing of such vehicles,[75]and directs that regulations be issued governing their operation.[76]

In 2012, Florida adopted similar legislation providing for the testing and operation of autonomous vehicles.[77]California also adopted legislation on the subject in 2012.[78]

According to the National Conference of State Legislatures, roughly 60% of states have adopted some form of legislation concerning autonomous vehicles.[79]

In some states, provisions for testing autonomous vehicles have been adopted through executive order or administrative regulation. In Arizona, the governor issued executive orders directing state agencies to devise regulations for the testing of autonomous vehicles.[80] This order was modified in 2018 after a pedestrian was fatally injured by an autonomous vehicle.[81] In this incident the National Transportation Safety Board found that a combination of the disabling of the vehicle’s automatic braking system and a failure to warn the on-board human operator of the presence of an unknown object led to the crash.[82]

Nevada began to issue regulations for the operation of autonomous vehicles in 2012.[83] California adopted regulations for the testing of autonomous vehicles in 2014. [84] The Division of Motor Vehicles adopted regulations for deployment of autonomous vehicles in 2018.[85]

According to the National Conference of State Legislatures, as of April 2017, twenty-eight states had introduced some form of regulations for autonomous vehicles.[86]

III. Other Government Actions Related to AI

A. Government Reports

The following are high-level federal government reports related to AI prepared by the current and previous administrations:

• EXECUTIVE OFFICE OF THE PRESIDENT, NATIONAL SCIENCE AND TECHNOLOGY COUNCIL, COMMITTEE ON ARTIFICIAL INTELLIGENCE, PREPARING FOR THE FUTURE OF ARTIFICIAL INTELLIGENCE (2016), https://obamawhitehouse.archives.gov/sites/default/files/whitehouse_files/microsites/ostp/NSTC/preparing_for_the_future_of_ai.pdf, archived at https://perma.cc/4RCY-PUS4. This report summarizes the status of AI in American society, economy, and government. It looks at specific applications, such as transportation and war fighting, and makes nonbinding recommendations.
• WHITE HOUSE OFFICE OF SCIENCE AND TECHNOLOGY POLICY, SUMMARY OF THE WHITE HOUSE SUMMIT ON ARTIFICIAL INTELLIGENCE FOR AMERICAN INDUSTRY (2018), https://www.whitehouse.gov/wp-content/uploads/2018/05/Summary-Report-of-White-House-AI-Summit.pdf, archived at https://perma.cc/Z7N7-24JE. This report reviews the Trump administration’s regulatory and policy approach to AI.

B. Position on Lethal Autonomous Weapons Systems

At the 2018 meeting of the GGE on LAWS, established under the Convention on Certain Conventional Weapons, the representatives of the United States argued that at this point it is too early for LAWS to be banned under the Convention. The delegation also presented its position that at this time it is not necessary to create a definition of LAWS, and that such systems should not be stigmatized, but instead  their use should be viewed within the framework of the current laws of war.[87]

Back to Top

Prepared by Eduardo Soares, Tariq Ahmad, Ruth Levush, Gustavo Guerra,
Senior Foreign Law Specialists
and
James Martin
Legal Information Analyst
January 2019

---