TLDR: The European Commission has created a clear division in the AI industry by establishing a voluntary Code of Practice for the AI Act and publishing the list of its initial signatories. Key industry players like Google and Microsoft have opted for cooperative compliance by signing, while the prominent refusal of Meta signals a strategy of resistance. This split effectively transforms the voluntary code into a de facto regulatory standard, forcing professionals in policy, ethics, and government to develop new, nuanced strategies for a fragmented AI governance landscape.
The European Commission has officially drawn a line in the sand for the artificial intelligence industry. By recognizing a voluntary Code of Practice as a key compliance tool for the AI Act and publishing the list of its initial signatories, the EU has done more than just create a procedural milestone. The resulting list of who has signed—and more importantly, who has not—is the clearest signal to date that the AI industry is fragmenting into distinct regulatory camps. For government, policy, and ethics professionals, this tactical release makes one thing clear: the era of monolithic AI governance is over, compelling a fundamental re-evaluation of engagement and oversight strategies.
The Great Divide: Decoding the Signatories and the Holdouts
The list of signatories reads like a who’s who of the AI world, with major players like Google, Microsoft, OpenAI, Amazon, and Anthropic all committing to the Code of Practice. These companies, despite voicing concerns that the regulations could stifle innovation, have chosen the path of cooperative compliance. Their calculus is clear: signing on offers a streamlined path to demonstrating compliance, increased legal certainty, and a more collaborative relationship with regulators. In essence, they are choosing to shape the regulatory future from within the system.
However, the most telling story comes from the most prominent non-signatory: Meta. The social media giant has explicitly refused to sign, with its Chief Global Affairs Officer labeling the code an “overreach” that introduces “legal uncertainties for model developers” and goes far beyond the scope of the AI Act itself. This isn’t merely a disagreement over fine print; it’s a fundamentally different philosophical stance. Meta’s decision, positioning the code as a direct threat to innovation, signals a strategy of resistance rather than cooperation and establishes a clear counterpoint to the collaborative camp.
From ‘Voluntary’ Code to De Facto Standard
To understand the depth of this fracture, one must appreciate the subtle power of this “voluntary” code. While not legally binding in itself, the Code of Practice—covering transparency, copyright, and safety—is designed as the primary tool for providers of general-purpose AI (GPAI) models to demonstrate compliance with the Act’s obligations which took effect on August 2. Signing the code is intended to reduce administrative burden and increase legal certainty. The EU AI Office has been explicit that it will offer closer collaboration to signatories and focus its enforcement activities on those who have not signed up. This approach effectively transforms the voluntary framework into a de facto regulatory benchmark. Companies now fall into two tiers: those presumed to be acting in good faith through their adherence to the code, and those who will inevitably face greater scrutiny and must prove their compliance through other, more burdensome means. For policymakers, this creates a two-track system for oversight before the first formal investigation is even launched.
A Fractured Landscape Demands a New Governance Playbook
This division is more than just corporate drama; it signals the maturation and fragmentation of the AI governance landscape, a trend that experts have been watching with caution. A one-size-fits-all approach to regulation is no longer viable. Professionals in the policy and ethics space must now adapt their strategies to this new reality:
- For Policymakers & Regulators: The divergence requires a more nuanced approach to oversight. Engagement with signatories like Google and Microsoft will likely revolve around monitoring their adherence to the code’s commitments. Conversely, interactions with non-signatories like Meta will require a more assertive, audit-focused strategy to ensure compliance with the AI Act’s core tenets. The Commission has already indicated it would take tougher action against companies that don’t sign up.
- For AI Ethicists & Safety Researchers: The split presents a crucial, real-world experiment. Researchers now have a clear basis for comparative analysis. Will the models developed by signatories demonstrate measurably different outcomes in safety, fairness, and transparency compared to those from non-signatories? This divide creates a natural laboratory for studying the impact of cooperative governance versus a more adversarial, compliance-driven approach.
- For Lobbyists & Public Affairs Specialists: The lines of engagement have been redrawn. The industry can no longer be addressed as a single entity. Strategies must be tailored to the distinct philosophies of these emerging camps. An argument that resonates with the “cooperative compliance” camp may be entirely ineffective with the “regulatory resistance” camp.
The Road Ahead: Navigating a Divided AI World
The single most important takeaway from the EU’s signatory list is that the AI industry’s regulatory adolescence is over. We are entering an era of defined factions, where companies are making firm strategic choices about their relationship with government and regulation. The next critical phase will be observing how the EU AI Office wields its enforcement power. Will it make an example of a non-signatory to give the Code of Practice teeth? And how will this fragmentation in Europe influence regulatory debates globally? For professionals tasked with navigating the intersection of technology, policy, and ethics, the message is clear: the path to comprehensive AI governance is no longer a single highway, but a series of diverging roads that demand more sophisticated and targeted navigation than ever before.
Also Read:
