AI Ethics and Accountability: Privacy, Safety, and Security in Artificial Intelligence

Name: AI Ethics and Accountability: Privacy, Safety, and Security in Artificial Intelligence
Author: Briana Brownell

Learn about privacy, safety, and security in artificial intelligence

Preview the first lesson free — get full access to all 6 lessons.

Request a demo Sign in

Course: On-Demand

(bright digital music) The United Nations General Assembly adopted the Universal Declaration of Human Rights in 1948. It's Article 12 reads, no one shall be subjected to arbitrary interference with its privacy, family, home, or correspondence, nor to attacks upon his honor and reputation. Everyone has the right to protection of the law against such interference or attacks. The right to privacy is increasingly important in our interconnected world where a myriad of highly sensitive information is digitized. This increasing digitization means there are ever more ways in which privacy rights can be violated. The European Union's General Data Protection Regulation, or GDPR, which came into effect in 2018, has influenced many privacy regulations around the world. The principles discussed in the GDPR have had a very strong impact on especially two principles, the consent and the control over the use of data. The principle of consent maintains that an individual should give permission or consent for their data to be used. This principle evolved from the concept of informed consent in the medical field where individuals are required to be informed about the risks, benefits and alternatives to proposed treatments. The definition of consent varies considerably from requirements to simply give notice of processing that provided data, rather than a way for individuals to explicitly opt in, to the ability to explicitly give consent, as well as the ability to revoke consent later. Those working in AI ethics should take time to clearly delineate what does and doesn't constitute consent, how that follows the relevant regulations, and what frameworks are in place to track and verify that consent was in fact given. A related principle is control over the use of data. This principle asserts that individuals should be able to control how their data is used. Some specific principles detail how this control could be exerted including, the ability to restrict processing, the right to rectification, and the right to erasure. This is given rise to a related principle, data portability. The principle of data portability requires that an individual should be able to transfer their own data to third parties. Several countries have developed regulations around data portability requiring companies to allow individuals to easily access and transfer data to competitors. This is considered a pro-competition regulation that levels the playing field and allows smaller companies to compete with large incumbents. As a result of the importance of privacy many organizations are adopting privacy by design. This directly integrates these rights and principles into the design of all their data collection mechanisms and all their AI system designs. Understanding the different principles involved and what actions organizations can take to support the protection of privacy is a first step in effectively protecting the privacy of stakeholders. Plus consent may relate to the contract between the entity providing the data and the entity gathering it. Personal data should only be used for the purpose of fulfilling an explicit or implicit contract with the entity holding the personal data. In other words, the entity holding personal data is not allowed to use the personal data for any other purpose. This principle is also related to the ability of the data provider to restrict processing. The GDPR has been very influential in recent years which can perhaps be attributed to the fact that the principles stated in it, particularly the concepts of consent, control, and portability are now standard parts of most international privacy regulations.

Intermediate Provider : Briana Brownell 6 Lessons · 13m in Arabic, German, English, Spanish, French, Portuguese, Chinese

Course Description

These lessons on Privacy, Safety, and Security in Artificial Intelligence will discuss General Data Protection Regulation, consent, data portability, the ability to restrict processing, right to rectification, and right to erasure. You will learn important concepts involving reliability and safety and discover how monitoring can detect potential risks and quickly mitigate harm if the system is not functioning as desired. These lessons will also cover how traceability is important to safety.

In discussing the broad array of concepts surrounding privacy, safety, and security in artificial intelligence, we’ll look at the value and influence of the European Union General Data Protection Regulation (GDPR) that came into effect in 2018. We’ll also review the link between reliability and safety. For a system to be safe, there must be a certain level of reliability in the performance and decisions made by the system. A risk management approach can be used to figure out what testing procedures are relevant to a given domain and determine risks that the system might pose to important considerations like privacy and human rights.

By the end of these lessons, you’ll learn that the security of AI systems is also related to unauthorized use of the data or the AI system. You’ll discover that unauthorized use can cause harm to the creators and the users of the system, which is why many organizations are considering security-by-design in the implementation of any AI system.

What You'll Learn

Understand the principles governing AI privacy, safety, and security, including the EU General Data Protection Regulation (GDPR) that came into effect in 2018
Explain how reliability and safety are interlinked in AI systems
Describe how monitoring can detect potential risks and quickly mitigate harm when a system is not functioning as desired
Assess the relevant security principles of an AI system, including issues of unauthorized use of data
Apply concepts of consent, data portability, restricting processing, right to rectification, and right to erasure
Examine testing procedures and a risk management approach for determining risks an AI system might pose

Key Takeaways

The EU General Data Protection Regulation (GDPR), which came into effect in 2018, addresses consent, data portability, the ability to restrict processing, the right to rectification, and the right to erasure.
For an AI system to be safe, there must be a certain level of reliability in the performance and decisions made by the system.
Monitoring can detect potential risks and quickly mitigate harm if a system is not functioning as desired, and traceability is important to safety.
A risk management approach can be used to determine relevant testing procedures and identify risks the system might pose to considerations like privacy and human rights.
Security of AI systems relates to unauthorized use of data or the system, which can harm creators and users, leading many organizations to consider security-by-design.

Frequently Asked Questions

What topics does this course cover?

It covers privacy, safety, and security in artificial intelligence, including the GDPR, consent, data portability, restricting processing, the right to rectification and erasure, reliability and safety, monitoring, traceability, testing procedures, and the security of data and outcomes.

What will I be able to do after completing this course?

You will be able to understand the principles governing AI privacy, safety, and security, explain how reliability and safety are interlinked, describe how monitoring can detect risks and mitigate harm, and assess the relevant security principles of an AI system.

What does the course say about the link between reliability and safety?

The course explains that for a system to be safe, there must be a certain level of reliability in the performance and decisions made by the system.

Why is security important for AI systems according to this course?

The security of AI systems is related to unauthorized use of the data or the AI system, which can cause harm to the creators and users, which is why many organizations consider security-by-design in implementation.

What lessons are included in this course?

The lessons are Consent and Control Over the Use of Data; Ability to Restrict Processing; Right to Rectification and Erasure; Reliability and Resilience of AI systems; Testing Procedures; and Security of Data and Outcomes.