Written by Tuija Marin – Data and accessibility professional, who drives on human-centric approach to data and AI utilization.  (09/2026)

Organizations increasingly rely on AI-driven analytics to improve decision-making, identify opportunities, and create competitive advantage. Yet AI systems are only as reliable as the data they are built upon. When data is incomplete, biased, or collected without adequate safeguards, AI can reinforce existing problems and generate misleading insights.

If you are a leader in an organization that wants to utilize the full potential of AI in your business analytics, this topic is particularly important for you.

From Compliance to Trust

As data has become a valuable asset in the digital era, and especially in the age of AI, the ethical implications of data collection and handling have become increasingly important. Public trust in data and AI depends on transparent, accountable, and responsible data practices, making ethical data collection not only a regulatory concern but also a societal necessity [1}. By establishing strong ethical foundations for data collection and handling, organizations can improve the reliability of analytics, strengthen stakeholder trust, and support more trustworthy AI outcomes.

While the EU has introduced regulations such as the GDPR and the AI Act to help ensure the ethical use of data, some practices remain legally valid while raising ethical concerns.

For example, an organization may provide information about its data collection practices that comply with legal requirements, yet the explanation may be so complex that individuals cannot realistically understand how much data is being collected, how it is being used, or how long it will be stored. In such cases, consent may be legally valid, but its ethical basis can reasonably be questioned.

So, if being compliant with regulation is a good start, but truly forward-facing organizations go above and beyond bare minimum.

Creating Trust Through Responsible Data Practices

By adopting ethical data collection principles, organizations can build trust, reduce reputational and legal risks, and, in some cases, gain a competitive advantage. As data becomes increasingly central to business operations and AI-driven innovation, ethical data practices play an important role in strengthening people’s confidence and supporting responsible decision-making.

So, what are the core principles of ethical data collection?

• Informed consent, as discussed earlier, consent is not meaningful if individuals do not understand what they are consenting to. It has been studied that while many consumers are aware of their privacy rights, they often have limited understanding of how emerging technologies affect their privacy. Another challenge is consent fatigue. People encounter privacy notices, cookie banners, and terms and conditions on an almost daily basis, making it difficult to carefully evaluate every request for their data.
• Transparency is closely related to informed consent, but it also extends to data lifecycle management, including how data is processed after it is collected and how it is secured. Research suggests that some people feel organizations can operate in a “gray zone” where certain practices may be legal but not necessarily moral or ethical, as laws can be perceived as vague and open to interpretation, leaving room for organizations to exploit loopholes. [2,3] In the era of AI, transparency has become especially important. As most people are not AI experts, they may have limited understanding of how their data can be used to support automated decision-making, profiling, recommendation systems, or generative AI applications. This highlights the importance of providing clear and easy-to-understand information about data sources, data usage, and how personal data contributes to analytics and AI-driven outcomes [4].
• Security is a foundational part of ethical data collection, as cyber-attacks are a daily threat in current digital environment and continue to evolve rapidly in the age of AI. Data protection should be built into every stage of the data lifecycle rather than treated as an afterthought. Organizations also need to consider the internal security of their data by evaluating who has access to it, how much data is actually needed, and what types of data are necessary for a specific purpose. When data collection purposes, access rights, and usage standards are clearly documented, the risk of unauthorized internal data use can be significantly reduced. As no system is completely fail-safe, organizations should prepare for security incidents before they occur by establishing clear procedures for responding to data breaches.
• Data minimization is an integral part of both a strong security culture and ethical data collection, as well as a legal requirement under the GDPR. It is easy to assume that collecting more data will automatically lead to better analytics, but organizations should carefully weigh the potential benefits against the associated risks. More data brings greater responsibility and increases exposure to risks such as privacy breaches, data leaks, and the misuse of personal information. Ethical data collection is not about gathering as much data as possible, but about collecting the right data for a clearly defined purpose.

Understanding Bias in Data

People often assume that data is objective. However, this is a myth.

Data reflects the decisions, processes, and circumstances under which it was collected, meaning that bias can be present long before analytics or AI models are applied. If these biases are not recognized and addressed, they can lead to inaccurate insights, unfair outcomes, and poor decision-making. One of the most effective ways to minimize bias in analytics is to understand the most common types of bias and how they can affect the data we use.

• Representation bias occurs when certain groups are underrepresented in a dataset compared to their presence in the real world. In other words, some people or experiences may be missing from the data altogether. For example, if lending analytics are based primarily on the records of people who have previously received loans, a system may reject an application from a younger person who has recently moved into the country due to their limited credit history, even if they have stable employment and no history of missed payments. Ethical data collection and handling require organizations to be aware of who may be underrepresented in their data and, where appropriate, take steps to address those gaps.
• Historical bias appears when historical inequalities, discrimination, or social structures are reflected in data, even when the data itself is factually correct and accurately recorded. The data may accurately describe what happened in the past, but what happened was not necessarily fair or ethical. For example, a hiring algorithm trained on historical recruitment data may favor one gender over another in professions that have traditionally been dominated by a particular gender. Without careful consideration, AI systems can reproduce and reinforce these historical patterns.
• Sampling bias occurs when data collection methods favor certain groups, perspectives, or behaviors over others. For example, if an organization collects customer feedback exclusively through a mobile application, the data represents the views of people who use the app rather than the entire customer base. As a result, the organization may draw inaccurate conclusions because the perspectives of individuals who do not use the app are excluded from the analysis.

Finally

As organizations increasingly rely on data and AI, ethical data collection has become both a business and societal necessity. Compliance with regulations such as the GDPR and AI Act provides an important foundation, but truly responsible organizations go beyond legal requirements. By prioritizing informed consent, transparency, security, data minimization, and bias awareness, organizations can build trust, improve the quality of their analytics, and support more trustworthy AI outcomes. Ultimately, ethical data practices are not a barrier to innovation but a foundation for sustainable and responsible use of data and AI.

Ultimately, trustworthy AI starts with trustworthy data.

Sources:

[1] (PDF) ETHICAL CONSIDERATIONS IN DATA COLLECTION AND ANALYSIS: A REVIEW: INVESTIGATING ETHICAL PRACTICES AND CHALLENGES IN MODERN DATA COLLECTION AND ANALYSIS. https://www.researchgate.net/publication/378789304_ETHICAL_CONSIDERATIONS_IN_DATA_COLLECTION_AND_ANALYSIS_A_REVIEW_INVESTIGATING_ETHICAL_PRACTICES_AND_CHALLENGES_IN_MODERN_DATA_COLLECTION_AND_ANALYSIS

[2] Understanding the Ethics of Data Collection and Responsible Data Usage. https://www.ucumberlands.edu/blog/understanding-the-ethics-of-data-collection

[3] “Whether it’s moral is a whole other story”: Consumer perspectives on privacy regulations and corporate data practices. https://www.usenix.org/system/files/soups2021-zhang-kennedy.pdf

[4] Transparency and explainability (OECD AI Principle) – OECD.AI. https://oecd.ai/en/dashboards/ai-principles/P7