Hook
A quiet disclosure from Crypto Briefing—a publication I’ve tracked for years—caught my attention last week. It wasn’t about a new token or a DeFi hack. It was about a person: Cami Clark, described as a “key advisor” to Anthropic CEO Dario Amodei, wielding “non-formal influence” over the AI company’s strategic decisions and capital acquisition. Two sentences, no background, no context. Yet for anyone who lived through the ICO mania of 2017 or the DeFi yield farming frenzy of 2020, the pattern is painfully familiar.
We burned out trying to own the future. And now, the same kind of opaque, relationship-driven power is shaping the companies that will define the next decade of AI—and by extension, the crypto infrastructure that AI will run on.
Context
Anthropic is the poster child for “safe AI.” Its flagship model Claude 3.5 consistently rivals GPT-4o, and its “Constitutional AI” approach has won trust from regulators and enterprises alike. The company has raised over $7 billion from Amazon, Google, and others, positioning itself as a direct competitor to OpenAI. But beneath the polished narrative of mission-driven governance lies a structure that critics call a “shadow decision layer.”
Anthropic is structured as a Public Benefit Corporation, with a Long-Term Benefit Trust designed to ensure its decisions align with human welfare. In theory, this is a model of transparency. In practice, the presence of a non-formal advisor like Cami Clark—whose role is not publicly disclosed, whose influence is not formally bounded, and whose accountability is undefined—echoes the very governance deficits that have plagued decentralized protocols for years.
We burned out trying to own the future, only to realize that the most critical decisions are often made in rooms without permanent doors.
Core
The analysis of the original article reveals a stark truth: the information density is near zero. We know only that Clark helps “secure key investments” and “shape strategic decisions.” No details on her background, her specific contributions, or whether her role is acknowledged by Anthropic’s board. Yet from this tiny data point, we can extrapolate a systemic risk that the crypto community should care about deeply.
First, the ethical dimension. In AI companies, informal influence creates an “accountability gap.” A formal CEO or board member is legally and publicly responsible for their decisions. A non-formal advisor is not—yet they can redirect strategy, open capital doors, and tilt the weight of a billion-dollar entity. This is the same governance flaw that led to the collapse of FTX: a small, unaccountable group of insiders making decisions that affected millions. The difference is that Anthropic is a for-profit PBC with an explicit safety mission, making the gap even more dangerous.
Second, the investment angle. The article’s author notes that Clark’s role is “indirectly catalytic” to commercialisation. In the AI race, capital access is a competitive moat. If Clark’s network (possibly linked to crypto/Web3, given the article’s source) helped Anthropic secure funding from Amazon or Google, that’s a strategic asset. But it also raises conflict-of-interest questions: What if Clark has ties to other portfolio companies? What if her advice steers Anthropic toward partnerships that benefit her network rather than the public interest? In crypto, we call this “insider privilege.” In AI, it’s called “advisory leverage.”
Third, the industry pattern. The analysis rightly points out that this isn’t unique to Anthropic. Sam Altman’s orbit, Demis Hassabis’s inner circle, and even Vitalik Buterin’s “informal council” in Ethereum all exhibit similar dynamics. The AI industry, like crypto, thrives on speed and trust. But speed without transparency breeds fragility. When I audited whitepapers during the 2017 ICO boom, I saw the same pattern: projects with a charismatic founder and a hidden network of advisors often raised the most money—and often failed the hardest.
We burned out trying to own the future, but the future is being owned by people we don’t even know.
Contrarian
Before we demonize the shadow advisor, let’s consider the counter-intuitive angle. In a hyper-competitive landscape like AI, formal governance can be a straitjacket. Boards are slow, committees are politicked, and publicly disclosed roles can be poached. Informal advisors allow companies to move fast, tap specialised expertise, and maintain strategic flexibility. Cami Clark might be exactly the reason Anthropic recovered from the FTX collapse and secured critical funding—because she could operate outside the bureaucratic noise.
Moreover, the crypto-native concept of “trustless systems” is often an illusion. Even the most decentralized DAOs rely on a small group of core contributors and influencers who steer governance. The real question is not whether informal influence exists, but whether it is disclosed and bounded. If Anthropic disclosed Clark’s role and scope, it would set a precedent. If not, it’s a governance time bomb.
The contrarian take: We should welcome the efficiency of informal networks, but demand a new standard of “shadow transparency.” Just as DeFi protocols now require smart contract audits, AI companies should require “influence audits”—a public disclosure of all non-formal advisors, their compensation, and the decisions they influence. That would preserve agility while restoring accountability.
Takeaway
The Cami Clark story is a microcosm of a much larger trend: the convergence of AI and crypto governance deficits. As AI models become the backbone of financial infrastructure, the same opaque relationships that shape Anthropic’s funding will shape the rules of the next generation of decentralized markets. The crypto community, which has learned painful lessons about trust and transparency, should be the first to demand clear disclosure of who really holds the strings.
Will the next AI breakthrough be shaped by a formal board, or by a quiet advisor with a Rolodex? The answer will determine not just the future of Anthropic, but the future of the decentralized economy that depends on it.
