From regulation to innovation: How certification can build trusted AI for a sustainable future

For over a decade, the narrative surrounding artificial intelligence (AI) has been one of speed: deploy quickly, scale rapidly and gain a first-mover advantage. Now, the year 2025 is a turning point for AI. The European Union’s (EU) AI Act has entered into force; meanwhile, the United States and Asia are advancing their own frameworks.

For many leaders, regulation is instinctively viewed as a cost or a hindrance to innovation. In fact, the opposite is true. Regulation is becoming a catalyst for trusted adoption, offering companies that embrace compliance early not just protection against fines but a competitive advantage in credibility, access and market share.

Just as the EU’s General Data Protection Regulation (GDPR) reshaped global cloud adoption, the AI Act and its international counterparts will define who earns trust – and who is left behind.

Multiple surveys show compliance and risk concerns are stalling AI: 77% of executives say regulatory uncertainty impacts decisions, while 74% paused at least one AI project in the last year due to risk.

The EU AI Act classifies systems into unacceptable (prohibited), high-risk (subject to assessment), limited-risk (transparency obligations) and minimal-risk (no obligations).

High-risk systems, in healthcare, transport, energy, or education, must undergo a conformity assessment before entering the market. Without this assurance, adoption stalls. With it, buyers, from hospitals to governments, can adopt AI solutions with confidence.

Too often, compliance is treated as a late-stage hurdle, bolted on after innovation. However, leaders who flip the model can make compliance a design driver. We call this the compliance-driven innovation loop:

Rather than slowing things down, this approach should accelerate adoption by eliminating the friction of uncertainty.

There are several use cases across different industries of validating AI applications:

Recent studies demonstrate how explainability tools enable physicians to understand why AI models classify skin lesions as malignant or benign. Meanwhile, reliability audits assess how consistently these systems perform under real-world conditions using metrics such as the Brier score.

Together, such methods demonstrate how certification frameworks can transform medical AI into solutions that doctors can trust, regulators can approve and patients can rely on.

Mercedes-Benz applied “compliance-by-design” in developing its Drive Pilot system. By embedding explainability and human-in-the-loop safeguards from the start and working with German regulators early, the company secured approval for Level 3 automated driving at 95 km/h.

This positions it ahead of competitors and opens procurement opportunities with fleet buyers who prioritize certification readiness.

Industrial projects such as ZeroDefectWeld show that AI can detect and classify weld defects on radiographs, reducing manual inspection error in industrial environments.

Grounding these systems in the EU AI Act – meeting Article 15 requirements on accuracy, robustness and cybersecurity and applying high-risk controls when the AI serves as a safety component – creates a clear, auditable path to compliant, AI-enabled Non-Destructive Testing across infrastructure projects.

The result: safer builds, faster delivery and more reliable assets, which is direct progress towards Sustainable Development Goal (SDG) 9 on infrastructure.

Microsoft is adapting its products and contracts to comply with the EU AI Act, updating policies to ban prohibited uses such as social scoring and signing the EU AI Pact.

It supports customers with Trust Centre documentation, transparency notes and governance tools such as Purview Compliance Manager and Azure AI Content Safety. By combining internal standards with regulatory engagement in Europe, Microsoft aims to help enterprises innovate with AI while staying compliant.

Across these cases, certification transforms regulation from a constraint into an enabler of scale.

Economically, investors are applying a “trust premium” to companies with strong governance. Procurement teams in government and critical infrastructure now demand conformity assessments upfront.

Socially, certification safeguards fundamental rights and helps AI align with the SDGs:

Politically, certification bridges high-level regulation with technical methods, enabling governments to harmonize standards across borders, thereby reducing fragmentation and facilitating global AI trade.

For executives, policy-makers and innovators, the agenda is clear:

Regulation clarifies the rules of the game, certification translates those rules into practice –together, they make AI not only powerful but trustworthy.

The leaders of tomorrow will not simply deploy advanced AI. They will deploy trusted AI by design, earning both market access and societal license to operate.