The European AI Act, the Digital Omnibus and the reality Europe can no longer ignore

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

I. A starting point Europe has tried to gloss over

In 2024 Europe celebrated the adoption of the EU Artificial Intelligence Act (Regulation (EU) 2024/1689) – the AI Act – as though it were a watershed moment. The narrative was simple: ethical leadership, protection of citizens and a regulation the world would admire.

The reality has been rather different. While the speeches were being made:

• The EU attracted just 10% of global AI investment in 2024–2025.
• The US accounted for 65%.
• Asia, for 25%.
• European start-ups in critical sectors – health, fintech, automation – moved straight to the US so that they could operate without obstacles.
• Mistral AI, one of the few European-born companies with global ambitions, opened an office in San Francisco in order to grow.

This is not a question of whether or not to ‘comply’. It is a design problem: the framework created to protect has built a wall that only Big Tech can climb.

Which brings us to the second piece on the board.

II. Before fixing the AI Act, Europe has already set another process in motion: the Digital Omnibus

In late 2025 the European Commission put forward what it calls the Digital Omnibus: a package of coordinated reforms to ‘simplify’ the European digital ecosystem, including:

• adjustments to the AI Act,
• amendments to the GDPR,
• a review of the ePrivacy rules,
• alignment with DORA,
• and harmonisation with MiCA and the DMA.

Its formal presentation is scheduled for 19 November 2025, although much of the content is already circulating in drafts and leaks.

The official message is ‘simplification’. The political effect is something else: an admission that Europe’s current digital framework does not work coherently. And this confirms what the market has sensed for a year.

III. Europe does not have a one-off problem: it has four simultaneous fault lines

1. Regulatory fault line: too many rules, no coordination

The AI Act, the GDPR, MiCA, the DMA, DORA and sector-specific regulation form a system whose parts were never designed to coexist.

Clear examples:

AI-assisted medical diagnosis – The AI Act requires large, representative datasets. – The GDPR requires strict data minimisation. The company must maximise and minimise at the same time.

Biometrics The same fingerprint may count as highly sensitive data or fall outside ‘high risk’ altogether, depending on how it is used.

Finance – AI Act: high risk. – GDPR: mandatory human intervention. – DMA: interoperability. – Financial supervisors: statistical validation.

The European Parliament has already acknowledged it: ‘lack of coordination, duplicated costs and contradictory criteria’.

The Digital Omnibus tries to correct part of the problem, but exposes a bigger one: Europe legislated before building an institutional architecture capable of sustaining that level of regulation.

2. Economic fault line: European costs that make it impossible to compete

• Compliance with the AI Act for a high-risk system costs €500,000 to €2 million.
• Aggregate impact on the EU in 2025: 10.9 billion.
• Result: 90% of the European GPAI market in the hands of US companies.

The Digital Omnibus does not remove this imbalance. It merely proposes partial, temporary relief for mid-caps and start-ups. It does not resolve the underlying asymmetry.

3. Timing fault line: timetables out of step with reality

• High-risk obligations: August 2026.
• Technical standards: Q2 2026, at best.
• Notified bodies: most not yet designated.
• Regulatory guidance arriving after the rules have entered into force.

The Digital Omnibus provides for delays and ‘clock-stops’. Not because Europe wants to be flexible, but because it cannot get there in time.

4. Global fault line: Europe competes alone

• US: light-touch regulation and rapid adaptation.
• China: strict control + 50 billion in subsidies.
• Europe: rigid regulation + no equivalent subsidies.

Foundation models in 2025: – Europe: 3 – US: 40 – China: 15

This is not a cyclical problem: it is structural.

IV. The AI Act relaxation reported in the press is not a reform: it is an admission

The leaks (FT, Reuters, Tech Policy Press) point to:

• delayed penalties,
• exemptions for medium-sized companies,
• relaxed GPAI rules,
• a clock-stop for high-risk systems.

The Digital Omnibus is the formal vehicle for introducing these adjustments. But its origin is clear: Europe has realised that the regulation, as it stands, is not sustainable in practice. Uncertainty, however, keeps growing: every leak causes paralysis. And the absence of clear political communication heightens the sense of improvisation.

V. What really works in 2025–2028?

1. Smart defensive classification Preventing systems from ‘falling’ into the high-risk category by default.
2. Avoiding open source without contractual traceability Especially until 2027.
3. Automated documentation from CI/CD Not optional: it is the only way to withstand simultaneous audits.
4. Active use of national sandboxes Spain, Germany and France are acting as regulatory decompression zones.
5. Honest communication with investors

‘We are aligning evolving standards with documentation that is traceable from the pipeline’. In an unstable environment, this message builds trust.

VI. Who is executing rather than improvising

This point matters.

Few organisations work at the genuine intersection of the legal, the technical and the operational. META Channel Corporation is one of them.

While others divide regulations up by department, META Channel Corporation translates the AI Act, the GDPR, MiCA, DORA and sector-specific regulation into a single operating model.

This makes possible something most organisations do not even know where to begin with: making five regulatory frameworks work as if they were one, without internal contradictions.

It is neither theory nor marketing. It is architecture. And it is what makes it possible to keep building while others are still debating interpretations.

VII. Europe needs clarity, not more layers of regulation

The AI Act was born of a legitimate political conviction. The Digital Omnibus is born of an urgent institutional need. But a continent cannot stand on principles alone. It needs coherent rules, realistic timetables and regulators who talk to each other.

If the Digital Omnibus amounts to minor adjustments rather than an in-depth reform, Europe will prolong its uncertainty until the end of the decade. Meanwhile, businesses cannot wait for Brussels to get its house in order.

The only sensible strategy is to prepare for the most realistic scenario: an AI Act that is partly corrected, partly watered down and applied by authorities that will go on interpreting it unevenly for years.

Those who adapt to that scenario will be able to compete. Those who do not will be trapped between rules that do not fit together.