Ethical and regulatory challenges in the age of autonomous agents (Chapter 2)

By Antonio Tejeda Encinas, President of the Euro-American Committee of Digital Law - CEA Digital Law, CEO of Meta Channel Corp

Introduction: a necessary sequel

In our previous article, entitled ‘Una Revolución Silenciosa en el Derecho Digital’, we explored how autonomous agents are transforming key areas of digital law. However, we left several fundamental questions open, such as how to allocate liability, how to mitigate bias and how to protect privacy and security in an increasingly automated world.

In this new analysis, we address these questions from an ethical and regulatory perspective, supporting our reflections with concrete examples that, as of December 2024, show the challenges these systems face in practice. This article seeks not only to answer the questions raised but also to lay the groundwork for future solutions that will guide the integration of autonomous agents within a fair and responsible framework.

Legal liability: a framework for autonomous decisions

Allocating liability is perhaps one of the greatest legal challenges in the age of autonomous agents. These systems, which operate independently and adaptively, can make decisions with significant consequences, yet they have no legal personality, which makes it difficult to attribute liability clearly.

Who is liable when things go wrong?

In 2024, a landmark case exposed just how complex this question is. An autonomous agent used by an investment firm in the financial sector made decisions based on biased historical patterns, causing losses running into the millions. The legal debate that followed asked whether liability lay with the system’s developers or with the company that deployed it, or whether it should be regarded as a risk inherent in the use of autonomous technologies.

Towards a new legal approach

An appropriate regulatory framework should:

1. Recognise the legal status of autonomous agents: although they are not independent legal actors, they could be treated as advanced tools subject to specific oversight.

2. Introduce compulsory insurance: like liability insurance, this would ensure that victims of harm are compensated regardless of who is directly responsible.

3. Create ongoing assessment mechanisms: monitor the decisions of autonomous agents, especially in critical sectors such as finance and healthcare, in order to prevent significant failures.

Mitigating bias: fairness in automated decision-making

Autonomous agents rely on historical data to learn and operate, but that data may reflect pre-existing prejudices and inequalities. In 2024, concrete examples show how these biases continue to affect artificial intelligence systems:

1. Virtual assistants such as Siri and Alexa: these systems still use female voices by default, reinforcing gender stereotypes and responding submissively to sexist commands or comments.

2. Recruitment processes: tools similar to the one Amazon withdrew in 2018 are still in use, showing bias in favour of certain genders and demographic groups.

3. The US justice system: predictive algorithms still overestimate the risk of reoffending for Black defendants and underestimate it for white defendants, perpetuating inequalities in criminal justice.

Strategies to reduce bias

1. Regular audits: systematic reviews to identify and correct bias in algorithms and data.

2. Algorithmic transparency: requiring clear explanations of how systems reach their decisions.

3. Diverse data: ensuring that training data fairly represents all population groups.

Privacy and security: safeguards in the age of big data

Autonomous agents process enormous volumes of personal data, which makes them a prime target for cyberattacks and raises ethical questions about how privacy is handled.

Concrete examples from 2024

1. Automated customer service in tourism: AI agents have shown bias in recommendations based on historical booking patterns, discriminating against certain groups of users.

2. Risk of sensitive data leaks: in healthcare and financial systems, cyberattacks have compromised the integrity of personal data handled by autonomous agents.

Regulatory proposals

1. Strict regulatory compliance: ensuring that agents comply with regulations such as the GDPR and with international standards.

2. Protection through advanced encryption: applying end-to-end encryption to data processed by autonomous agents.

3. Specialist oversight: setting up regulatory bodies to monitor their use in sensitive sectors.

A comprehensive ethical and regulatory framework

Meeting these challenges requires a regulatory framework that not only governs autonomous agents but also promotes their ethical development. This includes:

Ethics by design: building values such as fairness and transparency into the early stages of development.

Interdisciplinary collaboration: involving technologists, lawyers and sociologists in the design of public policy.

User training: equipping those who deploy these technologies to understand the risks and responsibilities involved.

Conclusion: balancing innovation and responsibility

The concrete examples from 2024 show that autonomous agents are already shaping critical sectors such as finance, justice and customer service. They also show how urgently we need to respond to the ethical and regulatory challenges they pose.

In this new phase of digital law, the challenge is not only to understand how these systems work but to decide how we want to integrate them into our society. A robust and adaptable regulatory framework will ensure that autonomous agents are not merely useful tools but also responsible and fair in their dealings with people. In this way, we will build a digital ecosystem that does not simply regulate the future but shapes it with justice and equity.