AI Governance Is Not Optional: What the EU AI Act Omnibus Means for Monitoring & Evaluation

Recent discussions about the EU Artificial Intelligence Act and the proposed “Digital Omnibus” adjustments have circulated widely across professional networks. Some commentaries suggested that organizations should slow down their AI governance efforts and wait for greater regulatory clarity. This interpretation risks creating more confusion than clarity.

As AI tools increasingly influence data collection, analysis, reporting, and evidence-based decision-making, evaluation systems must ensure transparency, accountability, and methodological rigor. The EU AI regulatory framework and the newly proposed Digital Omnibus adjustments aim to simplify compliance but leave the core governance obligations intact.

The AI Act’s Risk-Based Approach

The EU Artificial Intelligence Act introduces a risk-based regulatory framework for AI systems. Rather than regulating all AI applications equally, the Act focuses its strictest obligations on high-risk systems. High-risk AI systems typically include applications used in areas such as:

• recruitment and employee evaluation
• credit scoring and financial decision-making
• medical devices and healthcare diagnostics
• decision-support systems in critical infrastructure
• certain AI applications in education or public administration

Estimates suggest that only around 10–15% of AI systems fall into the high-risk category. Most AI tools currently used in organizations—including generative AI assistants, analytics tools, and productivity features integrated into software platforms—generally fall into lower risk categories. However, lower risk does not mean zero responsibility. Organizations still need to understand how AI systems influence decisions, data interpretation, and program outcomes.

The Real Governance Challenge: Visibility

In practice, many organizations face a challenge that has little to do with regulation itself. The challenge is visibility. When institutions begin mapping their internal systems, they often discover a fragmented landscape of AI use:

• AI tools adopted independently by departments
• generative AI features embedded in existing software platforms
• external AI services integrated into workflows
• automated decision-support systems operating without formal oversight

In this sense, the AI Act does not suddenly create the need for AI governance. Rather, it highlights the fact that many organizations never structured governance around AI technologies in the first place.

Why This Matters for Monitoring & Evaluation

Artificial intelligence is rapidly becoming part of the evaluation lifecycle. Many M&E teams are already using AI tools for tasks such as:

• automated survey analysis
• qualitative data coding
• document and report summarization
• predictive analytics in program design
• dashboard generation and real-time monitoring
• natural language processing for large datasets

These technologies can improve efficiency and expand analytical capacity. However, they also introduce new methodological and ethical considerations. For evaluators, three governance questions are becoming increasingly important.

1. Transparency in AI-Assisted Analysis

If AI tools assist with coding interviews, analyzing survey responses, or identifying patterns in qualitative data, evaluation teams must ensure methodological transparency. This means documenting: which tools were used, how outputs were generated, and what human verification processes were applied. Without clear documentation, the credibility of evaluation findings may be weakened.

2. Bias and Fairness in Evaluation Data

AI models trained on historical datasets may reproduce existing biases. In the context of development and social impact programs, this can influence analyses related to gender equality, social inclusion, vulnerability assessments, and beneficiary targeting. Evaluators must therefore ensure that AI-supported analysis does not unintentionally reinforce existing inequalities.

3. Accountability for AI-Supported Decisions

AI systems are increasingly used to support project risk assessments, performance monitoring dashboards, program targeting models, and resource allocation decisions. These systems may influence strategic decisions about programs and funding. A critical governance question arises: who remains accountable when AI-supported tools influence evaluation insights or program decisions? Maintaining human oversight remains essential.

Practical Steps for M&E Organizations

Waiting for perfect regulatory clarity will not address the governance challenge. Instead, organizations can begin strengthening their governance structures now.

AI Governance Is Becoming Part of Evaluation Quality

Monitoring and Evaluation has long emphasized principles such as transparency, accountability, methodological rigor, and ethical practice. As AI becomes integrated into evaluation systems, these same principles must now extend to AI governance and oversight. The EU AI Act does not fundamentally change the purpose of evaluation. But it does raise expectations about how organizations manage the technologies that increasingly support evidence generation, analysis, and decision-making.

Organizations that begin building governance structures now will likely find the transition straightforward. Those that wait may discover that the real challenge was never the regulation itself—it was the absence of internal governance.

Sources and References

For readers interested in the regulatory context and policy developments mentioned in this article, the following official resources provide further information:

• European Commission – AI Regulatory Framework
• European Commission – Digital Omnibus Proposal
• European Parliament – AI Act Policy Briefing
• EvalCommunity – Monitoring & Evaluation Resources
• EvalCommunity Academy – Learning Hub