The Choice That Will Shape the Future of the AI Industry: Microsoft's and Meta's Contrasting Responses to EU Regulations

1. Introduction: The Gravity of "European Standards"

On July 19, 2025, Reuters reported that "Microsoft intends to sign the EU Code of Practice, while Meta refuses," causing a stir in the European digital policy community on social media. The AI Act, which the EU enacted in August 2024, is the world's first comprehensive AI regulation, with its core General Purpose AI (GPAI) provisions set to apply in August 2025. The code is positioned as a "runway" for this, allowing signing companies to conduct risk assessments and transparency reports a year in advance in exchange for "interpretation guidance."Digital Strategy

2. Provisions and Significance of the Code of Practice

The code was developed by 13 independent experts,

1. Data Transparency—Summary publication of training corpora, copyright clearance policies

2. Risk Management—Systemic risk assessment, red team exercises, major incident reporting

3. Safeguards—Cyber protection, misuse prevention measures, external audits
   —comprising 10 chapters. The EU has provided an incentive, stating that "signing will offer advantages in future compliance certification," but it is not legally binding.Digital Strategy

3. Microsoft's Aim: Leading Advantage through "Regulatory Cooperation"

Microsoft President Brad Smith stated, "Participating in the creation of international standards for responsible AI is in our interest," choosing "safe driving" to smoothly deploy Azure OpenAI Service and Copilot in the European market.

• Success Experience with GDPR Compliance: Early response led to growth in cloud projects.

• Projects for EU Governments: "Regulatory compliance" is an essential value in public procurement.

• Partner Ecosystem: Transparency becomes a weapon in joint research with European system integrators and universities.ETBrandEquity.com

4. Meta's Rejection: The Argument of "Uncertainty and Excess"

Meta's spokesperson commented, "The code does not provide legal certainty and hampers development speed." In their open model strategy, including Llama 4,

• the obligation for data summarization poses a "risk of third-party misunderstanding."

• Systemic risk assessment "hinders collaborative experiments with the research community."

• Copyright measures "conflict with U.S. fair use,"
  raising strong concerns. For Meta, which has faced over $1.7 billion in fines due to GDPR sanctions, "European regulation" is now a cost center.POLITICO

5. The Temperature on Social Media: Mixed Reactions

• Supporters (Researchers and Civic Groups)

  • "MS chose the social contract."

  • "Meta is once again buying time with 'transition period business.'"

• Critics (Developer Community)

  • "A trap to exclude open model forces."

  • "The EU is letting go of the 'innovation hub' itself."
    As a representative post, Reuters journalist Foo Yun Chee pointed out, "Meta's refusal was expected, but the political world does not hide its discomfort," while the investment information account @newsinvesting calmly analyzed, "The impact on stock prices is limited."X (formerly Twitter)X (formerly Twitter)

6. The Next Move of the European Commission

On the 18th, the EU published "Interim Guidelines for GPAI," presenting risk assessment and external audit formats. In case of violations, fines of up to 7% of global sales or €35 million may be imposed. While "not signing the code" does not equate to penalties, stricter scrutiny in practical operations is likely.Reuters

7. Expert Comments

• EU Legal Scholar Henna Virkkunen
  "The code is 'soft law,' but the presence or absence of 'good faith' will be questioned in future review processes."

• German Venture Capital BlueYard Capital Partner
  "The cost of transparency is heavy, but it's reasonable if considered as human risk insurance."

• Japan IPA AI Network WG
  "There is a high possibility that the European framework will become the international de facto standard. Japanese companies should benchmark it."

8. Ripple Effects on Global Regulatory Competition

The U.S. is aligning principles with the NIST RMF, and the G7 is doing so with the "Hiroshima AI Process," but detailed implementation remains undecided. China has mandated some elements in its "Interim Measures for Generative AI" in 2024. The risk of a "splinternet," where companies are forced to manage multiple versions of models for each region without uniform regulatory granularity, is becoming a reality.Silicon UK

9. The Future of Investment and Innovation

• VC Caution: Evaluation reduction considering regulatory compliance costs

• Talent Mobility: Explosive demand for compliance experts

• Impact on the OSS Ecosystem: While the publication of heavyweight models decreases, the diversification of lightweight models and toolsets progresses
  If Microsoft secures public projects with a "cooperative model," Meta's "open model upfront investment" may become difficult to recover.

10. Conclusion: The Choice as "Philosophy"

The contrasting decisions this time reflect a philosophical divergence beyond mere legal risk calculation,whether to view AI as social infrastructure or as pure technological competition.The EU is drawing the world standard under the banner of "transparency and accountability," and Microsoft's joining strengthens that gravity. On the other hand, Meta's opposition serves as a warning against the speed and diversity of innovation being overshadowed by regulation.
When the code's full implementation begins in August 2025, which strategy was wise will become visible through the evaluation of the market and society.