e-Discovery Team

by Ralph Losey

Published October 25, 2024

The future of Artificial Intelligence isn’t just on the horizon—it’s already transforming industries and reshaping how businesses operate. But with this rapid evolution comes new challenges. Ethical concerns, privacy risks, and potential regulatory pitfalls are just a few of the issues that organizations must navigate. That’s where the Organisation for Economic Co-operation and Development (OECD) comes in. To help groups embrace AI responsibly, the OECD has developed a set of guiding principles designed to ensure AI is implemented ethically and effectively. Are you prepared to harness the power of AI while safeguarding your company against the risks? Discover how the OECD’s blueprint can help guide you through this complex landscape.

Introduction

The Organisation for Economic Co-operation and Development (OECD) plays a vital role in shaping policies across the world to foster prosperity, equality, and sustainable development. In recent years, the OECD has shifted its focus toward the responsible development of AI, recognizing its potential to transform industries and economies. For businesses any other organizations considering the adoption of AI into their workflows the OECD’s AI Principles (as slightly amended 2/5/24) provide a good starting point to develop internal policies. They can help guide your board to make decisions that ensure AI technology is deployed ethically and responsibly. This can help protect them from liability, and their employees, customers, and the world from harm.

What is the OECD?

The Organisation for Economic Co-operation and Development (OECD) is an independent, international organization dedicated to shaping global economic policies that are based on individual freedoms and democratic values. The U.S. was one of the twenty founding members in 1960 when the Articles of the Convention were signed, establishing the OECD. It now has 38 member countries, mainly advanced economies. Though the OECD initially focused on economic growth, international trade, and education, it has become increasingly concerned with the ethical and responsible development of artificial intelligence.

In 2019, the OECD introduced its AI Principles–the first intergovernmental standard for AI use. These principles reflect a growing recognition that AI will play an important role in global economies, societies, and governance structures. The OECD’s mission is clear: AI technologies must not only drive innovation but also be applied in ways that respect human rights, democracy, and ethical principles. These AI guidelines are vital in a world where AI could be both a powerful tool for good and a source of significant risks if misused. The Five AI Principles and Recommendations were slightly amended on February 5, 2024.

The OECD is a highly respected group that collaborates with many international organizations, such as the United Nations (UN), World Bank, International Monetary Fund (IMF), and World Trade Organization (WTO). The OECD helps these groups align and coordinate efforts in global governance and policymaking. The OECD also engages in regional initiatives, providing tailored advice and support to specific regions such as Latin America, Southeast Asia, and Africa. Bottom line, the OECD has long played a crucial role in shaping global policy, promoting international cooperation, and providing data-driven, evidence-based recommendations to governments around the world.

Five Key OECD AI Principles

Before starting an AI program, businesses should consider the potential risks that AI poses to their operations, employees, and customers. By taking proactive steps to mitigate these risks, organizations can safeguard themselves from unforeseen consequences while reaping the benefits of AI. The OECD’s AI Principles (amended 2/5/24) represent one of many frameworks businesses should evaluate when integrating AI technologies into their operations. It is well respected around the world and should be a part of any organization’s due diligence.

These principles are built around five core guidelines:

Principle 1. Inclusive Growth, Sustainable Development, and Well-being

The first OECD AI principle stresses that AI should promote inclusive growth, sustainable development, and well-being for individuals and society. AI should benefit people and the planet. This core value reflects the potential of AI to contribute to human flourishing through better healthcare, education, and environmental sustainability.

Companies should be aware of the many challenges ahead. While AI-driven solutions, such as climate modeling or precision agriculture, can help tackle environmental crises, there is concern that rapid technological advancements may lead to widening inequality. For instance, the automation of jobs could disproportionately affect lower-income workers, potentially exacerbating inequality. Thus, this principle necessitates a strategy that ensures AI’s benefits are distributed equitably.

For businesses considering AI, three key actions should always be top-of-mind for board members:

• Engage Relevant Stakeholders: Before implementing AI, include a diverse group of stakeholders in the decision-making. This should involve executives, legal and data privacy experts, subject matter experts, human resources, and marketing/customer support teams. Each group brings unique perspectives that can help ensure the AI program is equitable and aligned with the company’s values.

• Evaluate Positive and Negative Outcomes: Consider both the potential benefits and risks to AI users and individuals whose data may be processed. AI should enhance productivity, but it must also respect the well-being of all involved parties.

• Consider Environmental Impact: AI systems require substantial computational resources, which contribute to a large carbon footprint. Sustainable AI practices should be considered to reduce energy consumption and minimize environmental impact.

Principle 2. Respect for the rule of law, human rights and democratic values, including fairness and privacy.

The wording of the second principle was revised somewhat in 2024. The full explanation for revised Principle Two is set out in the amendment recommendation of February 5, 2024.

a) AI actors should respect the rule of law, human rights, democratic and human-centred values throughout the AI system lifecycle. These include non-discrimination and equality, freedom, dignity, autonomy of individuals, privacy and data protection, diversity, fairness, social justice, and internationally recognised labour rights. This also includes addressing misinformation and disinformation amplified by AI, while respecting freedom of expression and other rights and freedoms protected by applicable international law.

b) To this end, AI actors should implement mechanisms and safeguards, such as capacity for human agency and oversight, including to address risks arising from uses outside of intended purpose, intentional misuse, or unintentional misuse in a manner appropriate to the context and consistent with the state of the art.

Respecting human rights means ensuring that Generative AI systems do not reinforce biases or violate individuals’ rights. For example, there is growing concern over the use of AI in facial recognition technology, where misidentification disproportionately affects marginalized groups. AI must be designed to avoid such outcomes by integrating fairness into algorithms and maintaining democratic values like transparency and fairness.

Businesses integrating AI into their operations should address several legal issues, including intellectual property, data protection, and human rights laws. To do this there are four things a board of directors should consider:

• Ensure Compliance with Laws: Verify that  Generative AI (GAI) adheres to copyright laws and data protection regulations such as GDPR or CCPA. Implement safeguards to ensure the system does not infringe upon users’ privacy or autonomy.

• Prevent Discrimination: Conduct thorough audits to ensure that GAI outputs are fair and free from discrimination. Discriminatory outcomes can damage reputations and result in legal challenges.

• Monitor for Misinformation: GAI systems must be designed to resist distortion by misinformation or disinformation. Mechanisms should be in place to quickly halt GAI operations if harmful behaviors are detected.

• Develop Policies and Oversight: Establish clear policies and procedures that govern the use of GAI within your business. This includes implementing human oversight to ensure AI actions align with ethical and legal standards.

Principle 3. Transparency and Explainability

Transparency and explainability are fundamental to user trust in AI systems. This principle calls for AI systems to be transparent so that users can understand how decisions are made. With complex AI algorithms, it is often difficult to decipher how certain outcomes are generated—a problem referred to as the “black box” issue in AI.

While transparency enables users to scrutinize AI decisions, the challenge lies in making these highly technical systems comprehensible to non-experts. This requires a good education program by experts. Moreover, explainability must strike a balance between safeguarding intellectual property and providing adequate insight into AI operations, especially when used in public sector decision-making.

Businesses and other organizations must ensure that employees and other users of its computer systems understand when and how AI is used, along with some understanding of how AI decisions are made, and what mistakes to look out for. See e.g. Navigating the AI Frontier: Balancing Breakthroughs and Blind Spots (e-Discovery Team, October 2024). For businesses, ensuring transparency involves two critical steps:

• Inform Users: Be transparent with employees, consumers, and stakeholders that GAI is being used. Where required by law, obtain explicit consent from users before collecting or processing their data.

• Explain AI Processes: Provide clear, easy-to-understand explanations of how AI systems function. This includes offering insight into the sources of data used for training the AI and explaining the logic behind AI outputs, such as content recommendations or predictions. It is also important to explain the errors to look out for and other idiosyncrasies of the system to look out for. Everyone should be taught the “trust but verify” process and remember that they are ultimately responsible for their actions, not the AI. See e.g. Panel of AI Experts for Lawyers: Custom GPT Software Is Now Available (6/21/24); Can AI Really Save the Future? A Lawyer’s Take on Sam Altman’s Optimistic Vision (10/04/24).

Principle 4. Robustness, Security, and Safety

This principle demands that AI systems be resilient, secure, and reliable. As AI systems are increasingly integrated into sectors like healthcare, transportation, and critical infrastructure, their reliability is essential. A malfunctioning AI in these areas could result in dire consequences, from life-threatening medical errors to catastrophic failures in critical systems.

Cybersecurity is a significant concern, as more advanced AI systems become attractive targets for hackers. The OECD recognizes the importance of safeguarding AI systems and other systems from security breaches. All organizations today must guard against malicious attacks to protect their data and public safety. Organizations using AI must adopt a comprehensive set of IT security policies. Two key actions points that the Board should start with are:

• Plan for Contingencies: Implement a Cybersecurity Incident Response Plan that outlines steps to take if the AI or other technology system malfunctions or behaves in an undesirable manner. This plan should detail how to quickly halt operations, troubleshoot issues, and safely decommission the system if necessary. You should probably have legal specialists on call in case your systems are hacked.

• Ensure Security and Safety: Businesses should continuously monitor their technology and AI systems to ensure they operate securely and safely under various conditions. Regular audits, including red team testing, can help detect vulnerabilities before they become significant problems.

Principle 5. Accountability

Accountability in AI development and use is paramount. This principle asserts that those involved in creating, deploying, and managing AI systems must be held accountable for their impacts. Human oversight is critical to safeguard against mistakes, biases, or unintended consequences. This is another application of “trust but verify” on a management level. This is particularly relevant in scenarios where AI systems are set up to help make decisions affecting people’s lives, such as loan approvals, hiring decisions, or judicial sentencing. These should never be autonomous, but recommendation with a human in charge. This is especially true for physical security systems.

A clear accountability framework is critical. The accountability principle ensures that even in highly automated systems, human oversight is necessary to safeguard against mistakes, biases, or unintended consequences. The Board of Directors should, as a starting point:

• Designate Responsible Parties: Assign specific individuals or departments to oversee the AI system’s operations. These stakeholders must maintain comprehensive documentation, including data sets used for training, decisions made throughout the AI lifecycle, and records of how the system performs over time.

• Conduct Risk Assessments: Periodically evaluate the risks associated with AI, particularly in relation to the system’s outputs and decision-making processes. Regular assessments help ensure the system continues to function as intended and complies with ethical standards.

Strengths and Weaknesses of the OECD AI Principles

The OECD AI principles are ambitious and reflect a comprehensive effort to create a global framework for responsible AI. However, while these guidelines are strong, they are not without their weaknesses.

Strengths

• Comprehensive Ethical Guidelines: The principles cover a broad spectrum of ethical concerns, making them a strong foundation for policy guidance.

• Global Influence: As an international standard, the OECD AI Principles provide a respected baseline for countries worldwide, not just the U.S. This allows for a coordinated approach to AI governance.

• Commitment to Human Rights: By centering AI development on human dignity and rights, the OECD ensures that ethical concerns remain at the forefront of AI advancements.

Weaknesses

• Lack of Enforcement: One of the significant drawbacks is the absence of enforcement mechanisms. The principles serve as guidelines, but without penalties for non-compliance, their effectiveness could be limited. A Board should add appropriate procedures that track their existing policies.

• Ambiguity in Accountability: While the principle of accountability is emphasized, the specifics of assigning responsibility in complex AI systems remain unclear.

• Underdeveloped Consideration of AI Bias: Although fairness is mentioned, the principles lack detailed guidelines on mitigating algorithmic bias, a significant concern in many AI applications. See e.g. Worrying About Sycophantism: Why I again tweaked the custom GPT ‘Panel of AI Experts for Lawyers’ to add more barriers against sycophantism and bias (e-Discovery Team, 7/9/24); Stochastic Parrots: the hidden bias of large language model AI (e-Discovery Team, 3/25/24).

In addition to the OECD international Principles, businesses should consult other frameworks to strengthen their AI governance strategies. For example, the NIST-AI-600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (7/26/24) provides much more detailed, technical guidance into managing the risks associated with AI technologies. Organizations may also want to consider the U.S. Department of State Risk Management Profile for Artificial Intelligence and Human Rights. It states that it is intended as a practical guide for organizations to design, develop, deploy, use, and govern AI in a manner consistent with respect for international human rights.

Conclusion

Implementation of the OECD’s Five AI Principles is an essential step toward the responsible development of AI technologies. While the principles address key concerns such as human rights, transparency, and accountability, they also highlight the need for ongoing international collaboration and governance. In many countries outside of the U.S. there are, for instance. much stronger laws and regulations governing user privacy. Following the OECD Principles can help with regulatory compliance and show an organizations good faith to attempt to follow complex regulatory systems.

By relying on multiple AI frameworks, not just the OECD’s, businesses and their Boards can ensure a comprehensive approach to AI implementation. In the rapidly evolving field of AI, where state and foreign laws change rapidly, it is prudent for any CEO or Board of Directors to base it policies on stable, well-respected, principles. That can help establish good faith efforts to handle AI responsibly. Consultation with knowledgeable outside legal counsel is, of course, an important part of all corporate governance, including AI implementation.

Documenting Board decisions and tying them back to internationally accepted standards on AI is a good practice for any organization, local or global. It may not protect all of a company’s decisions from outside attack based on unfair 20/20 hindsight, but it should provide a solid foundation for good faith based defenses. This is especially true if these principles are adopted proactively and implemented with advice from respected third-party advisors. We are facing rapidly changing times, with both great opportunities and dangers. We all need to make our best efforts to act in a responsible manner and the OECD principles can help us to do that.

Click here to listen to an AI generated Podcast discussing the material in this article.

Ralph Losey Copyright 2024 — All Rights Reserved