The most dangerous assumption in Washington right now is that technology companies can govern themselves. Not through malice. Not through conspiracy. Through the quiet, patient accumulation of influence in spaces where oversight was never designed to reach.
Consider a pattern I observed during my years reviewing smart contract architectures. The same structural vulnerability appears again and again: when the entity responsible for enforcing rules also benefits from their interpretation, compliance becomes a performance rather than a commitment. Auditors call this a conflict of interest. Politicians call it lobbying. The result is identical: the public bears the risk while private actors capture the reward.
The reported appointment of Jay Carney—former White House Press Secretary under Obama, former Google senior advisor—to lead a new AI policy center backed by venture capitalist Ron Conway represents exactly this kind of structural vulnerability. It is a move that appears reasonable on its surface, even sophisticated, but conceals a fundamental contradiction at its core: a policy center designed to shape AI governance, funded and directed by individuals whose primary financial interest lies in the least restrictive possible regulatory environment.
This is not a story about Carney's qualifications, which are genuine. It is not a story about Conway's intentions, which are likely sincere. It is a story about structures—the invisible architecture that determines whether governance serves the governed or those who claim to speak for them.
The Translation Problem Nobody Wants to Acknowledge
To understand why this appointment matters, you must first understand the most underappreciated challenge in AI policy: the translation gap.
I spent three years embedded in Istanbul's blockchain development scene, watching talented engineers build protocols that were technically elegant but practically incomprehensible to regulators. The reverse was equally true—policy makers would propose rules that were conceptually sound but technically unworkable. The failure was never about intelligence or intent. It was about vocabulary. Two communities speaking adjacent languages, each convinced the other was being deliberately obtuse.
Carney's background addresses this gap in ways that pure academics or career bureaucrats cannot. He served as White House Press Secretary from 2011 to 2014, a role that demands fluency in two radically different registers: the precise, careful language of official statement and the interpretive, narrative-driven language of public communication. He then spent four years as Google's senior vice president of global affairs and communications, learning to translate corporate technical priorities into government-compatible language.
This bilingual capacity is genuinely rare. I have reviewed dozens of blockchain governance proposals where the fundamental failure was not technical merit but communicative incompetence—the inability to make technically sound arguments accessible to non-technical decision makers. Carney's resume suggests he does not suffer from this disability.
The question is what he will translate, and for whom.
The Conway Variable: Venture Capital's Quiet Revolution
Ron Conway occupies a unique position in Silicon Valley's hierarchy. Often called the "Godfather of Silicon Valley," his investment firm SV Angel has backed companies that now represent a substantial portion of the technology industry's market capitalization. His network extends across every major technology company, every meaningful startup, and increasingly, every conversation about technology policy that occurs in Washington.
Conway's reported decision to establish an AI policy center—and his selection of Carney as its leader—reveals a strategic calculation that deserves closer examination. He is not funding academic research. He is not supporting independent think tanks in the traditional sense. He is building an infrastructure node: a place where industry perspectives can be processed, refined, and delivered to policymakers in formats optimized for receptivity.
This is not inherently sinister. Advocacy is legitimate. Industry associations have always existed. The problem is structural: when the entity advocating for a particular regulatory outcome also funds the research that informs that outcome, and also selects the personnel who will present it, the distinction between information and influence dissolves.
I encountered this dynamic repeatedly during my audit work. Protocol foundations that claimed to be independent but whose primary funders sat on governance councils. Auditing firms that provided security assessments to clients who later became significant financial backers. The conflict was always present, even when everyone involved was acting in good faith. Structural incentives are more powerful than individual intentions.
The American Regulatory Vacuum and Its Consequences
To appreciate the significance of this development, you must understand the regulatory context in which it occurs.
The European Union enacted its AI Act in 2024, establishing a comprehensive framework for artificial intelligence governance that classifies systems by risk level and imposes corresponding obligations. The legislation passed after years of negotiation, during which technology companies lobbied extensively to shape its provisions. The result is a framework that is more restrictive than the United States currently requires but arguably more permissive than many safety researchers would prefer—a political compromise that reflects the distribution of power within the EU's institutional structure.
China has developed its own AI governance framework, focused on algorithmic recommendation systems, generative AI services, and synthetic media. The framework emphasizes content control and regulatory compliance in ways that reflect the Chinese government's priorities but also establishes clear rules of the road for domestic AI companies.
The United States, meanwhile, has produced no comprehensive federal AI legislation. The Biden Administration issued an Executive Order on AI in October 2023, but executive orders are revocable and limited in scope. The Congressional AI Caucus exists but has not produced legislation. State-level activity is fragmentary: Colorado passed an AI law in 2024, California is considering multiple proposals, and a dozen other states have active legislation at various stages.
This vacuum is not accidental. It reflects the success of an ongoing lobbying effort by technology companies to prevent the creation of binding federal rules. The industry has consistently argued for self-regulation, voluntary frameworks, and sector-specific guidance rather than comprehensive legislation. The strategy has worked: American AI companies currently operate under a combination of sector-specific regulations, voluntary commitments, and common-law liability—none of which provides the predictability that serious long-term investment requires.
Conway's policy center, if it materializes as reported, will insert itself into this vacuum with significant resources and connections. Carney's government background provides access. Conway's network provides legitimacy. Together, they represent a potential shift in the balance of power within AI policy discussions—not through legislation, but through the quieter mechanism of agenda-setting and narrative framing.
The Trust Deficit That Cannot Be Audited Away
During the 2022 bear market, I led risk assessment for a stablecoin protocol that was navigating the collapse of several competitors. The lesson I took from that experience was not about technical architecture or smart contract security. It was about trust architecture.
The protocol that survived—the one that retained user confidence through the crisis—was not the most technically sophisticated. It was the one whose governance structure was most legible, whose decision-making processes were most transparent, and whose conflicts of interest were most explicitly acknowledged. When users could see exactly who was making decisions, and why, and what they had to gain, they chose to stay. When users could not see those things, they left—even when the underlying technology was sound.
This is the challenge facing the Conway-Carney policy center: legitimacy requires transparency, but transparency reveals the structural problem at its core.
If the center is funded primarily by technology companies with direct interests in AI policy outcomes, its research will be viewed with suspicion by academic reviewers and government officials who have learned to discount industry-funded studies. If the center attempts to position itself as an objective voice, it will face credibility attacks from those who can trace its funding back to venture capital portfolios that benefit from permissive regulation.
The center cannot solve this problem by hiring credible people or publishing credible research, because the structural incentive will always be present, visible to anyone who looks. Trust, in this context, is not a feature that can be added. It is an archived receipt that must be earned through consistent behavior over time—and that archive must itself be trustworthy.
The Representative Paradox
There is a second structural problem that receives less attention but may prove more significant: the representative paradox.
If Conway's center represents Silicon Valley's AI industry interests—which, given its funding structure, seems likely—it faces a choice that has no good answer. Represent the largest companies (Google, Meta, OpenAI, Anthropic) and be dismissed as a captured tool of corporate interests. Represent the broad technology industry including smaller companies and the center may find itself advocating positions that its largest funders oppose. Attempt to represent the public interest and the center must explain why it is funded by parties with direct financial interests in outcomes.
I have seen this paradox play out in blockchain governance many times. Token foundations that claimed to represent "the community" while being funded primarily by early token holders. Protocol governance councils that claimed to be neutral while being composed entirely of large token holders. The result was always the same: governance structures that were formally inclusive but substantively captured by those with the greatest financial stakes.
Carney's selection as director suggests the center has chosen a particular answer to the representative paradox: position itself as a sophisticated intermediary between industry and government, translating between the two in ways that privilege certain industry perspectives while maintaining the appearance of neutrality. This is a defensible strategy. It is also, I would argue, a strategy that systematically advantages the largest players who can afford sophisticated policy representation, while disadvantaging smaller companies, academic researchers, and civil society organizations who cannot.
What Independent AI Governance Actually Requires
The question that should be asked about this development is not whether Carney is qualified or whether Conway's intentions are good. It is whether an industry-funded, industry-directed policy center can serve the public interest in AI governance—and the answer, based on everything I have observed about similar structures in the blockchain space, is probably not in the way that matters most.
Independent governance requires three things that this structure cannot easily provide: distance from those who benefit from particular outcomes, transparency about funding and influence, and accountability to those affected by decisions.
Distance requires separation between the entity studying problems and the entities with interests in solutions. This does not mean industry perspectives should be excluded from policy discussions—they contain genuine and valuable information—but it means they cannot be the primary voice shaping the research agenda, funding the research, and presenting the conclusions.
Transparency requires that the public can see exactly who is paying for research, what they expect to gain from it, and how the research was conducted. The blockchain industry has made some progress on this front through practices like public treasury disclosures and governance voting records, though significant gaps remain. AI policy research has not made comparable progress.
Accountability requires that entities making governance recommendations can be held responsible for their outcomes. Industry-funded centers can be defunded if they produce unwelcome conclusions. Academic institutions can lose accreditation. Government agencies can be reformed. The accountability mechanisms for industry-funded policy centers are weaker, which means they face weaker incentives to produce honest analysis.
The Global Context Nobody in Silicon Valley Wants to Discuss
There is a competing narrative about this development that deserves consideration: perhaps Conway and Carney are doing exactly what the United States needs to maintain its position in the global AI competition.
The EU AI Act is now law, with enforcement beginning in stages. China's AI governance framework is consolidating. Both regulatory environments impose constraints on AI development that American companies view as obstacles to innovation. If the United States does not develop a coherent policy response, American AI companies may find themselves at a competitive disadvantage—forced to comply with rules they had no voice in shaping, or excluded from markets that require compliance with foreign regulations.
From this perspective, Conway's center represents American industry organizing to defend its interests in a global regulatory competition. Carney's background suggests the center will advocate for a regulatory approach that balances safety concerns with innovation incentives, positioning the United States as a moderate alternative to both the EU's precautionary approach and China's state-directed model.
This is a coherent strategic vision. It is also, I would argue, a vision that systematically underweights the safety risks that independent researchers have identified as potentially catastrophic. The competitive framing assumes that AI development will continue in roughly its current direction and that regulatory differences will primarily affect the pace and distribution of benefits. It does not adequately account for scenarios in which certain AI capabilities prove genuinely dangerous and require constraints that reduce competitive advantage in the short term.
The history of other technologies suggests caution here. The pharmaceutical industry spent decades fighting FDA regulations that it argued would slow innovation and disadvantage American companies internationally. The resulting regulatory framework is imperfect, but it has also prevented numerous dangerous drugs from reaching patients. The financial industry spent decades fighting derivatives regulations that it argued would impair American competitiveness. The 2008 crisis demonstrated the costs of that regulatory failure.
AI may be different. It may also not be. The question of which risks are worth taking in pursuit of competitive advantage is not one that should be answered exclusively by those who gain from the advantage.
What This Means for the Future of AI Governance
If the reported appointment proceeds and the center launches as described, it will represent a significant escalation in the professionalization of AI policy advocacy. The center will not be the first industry-funded policy organization—others exist—but it will bring significant resources, connections, and sophistication to the task of shaping how AI is governed.
This development should prompt reflection by three audiences.
For policymakers in Washington: the arrival of a well-funded, well-connected policy center should strengthen incentives to develop genuine regulatory capacity within government itself. The executive branch needs technical staff who can evaluate industry claims. Congress needs expert advisors who are not simultaneously representing industry clients. The current regulatory vacuum does not reflect lack of interest; it reflects lack of capacity. Filling that capacity should be a priority regardless of what industry-funded centers do.
For the AI research community: the center represents a competitive challenge to academic institutions that have historically provided independent analysis. If academic researchers want their work to remain relevant to policy discussions, they need to develop the communicative capacity and policy relationships that Conway and Carney will bring. The translation gap I described earlier must be addressed by both sides.
For the public: this development should reinforce the importance of demanding transparency about who funds AI policy research and what they expect from it. Trust requires verification. In a domain as consequential as AI governance, the default should be skepticism about any research or advocacy that cannot clearly identify its funding sources and demonstrate structural independence from those who benefit from its conclusions.
History is the only consensus that never forks. The structures we build now to govern AI will shape its development for decades. Those structures will only be as trustworthy as the processes through which we create them. An image is fleeting; its hash is the truth. In this case, the hash has not yet been recorded—but the foundations are being laid, and we should be watching carefully to see what kind of structure they support.
The question is not whether Silicon Valley will have a voice in AI governance. It will. The question is whether that voice will be the only voice, or whether it will be one among many, constrained by accountability mechanisms that ensure it speaks for more than its own interests.
That question remains open. And the answer will be determined not by the quality of the arguments or the credentials of the advocates, but by the structural choices we make in the years ahead—choices about who gets to participate in governance, who pays for research, and who bears the consequences when governance fails.
Those are the choices that matter. Everything else is commentary.