A report published by the Washington Free Beacon is drawing fresh attention to the people responsible for shaping the ethical guidelines built into Claude, Anthropic's flagship AI assistant. The piece profiles several members of Anthropic's ethics and policy teams, highlighting publicly stated views on morality that some readers are finding unsettling. Among the quotes cited is one researcher who said plainly: "I don't think morality is objective." That statement, and others like it, are now fueling a broader debate about transparency, values alignment, and who exactly gets to define an AI's conscience.
The People Behind Claude's Values
Building an AI system that behaves well requires making countless judgment calls. What counts as harmful content? When should an AI push back on a user? What political or social topics require neutrality, and which ones have clear right answers? These questions don't resolve themselves algorithmically. Humans answer them, and those humans carry their own philosophical commitments. Anthropic employs a team of researchers, ethicists, and policy specialists whose work feeds directly into how Claude is trained to respond. The Free Beacon report attempts to put faces and views to that team, drawing on social media posts, academic writing, and public statements.
Key Facts
- The Free Beacon report profiles multiple members of Anthropic's ethics and trust teams by name.
- At least one researcher is quoted saying they do not believe morality is objective.
- Anthropic has published detailed documentation on Claude's values, known as the model spec, which describes a calibrated, empirical approach to ethics.
- The report is being cited by critics who argue AI companies lack ideological diversity in their ethics functions.
- Anthropic has not issued a public response to the specific claims in the piece.
The timing matters. Anthropic is expanding rapidly, signing major enterprise deals and pushing Claude's model family into sensitive domains including healthcare and scientific research. The ethics teams profiled in the report are not peripheral figures. Their decisions have real consequences for millions of users. Whether those decisions reflect a coherent and defensible philosophical framework, or simply the personal views of a small group of researchers, is a legitimate question worth examining.
"I don't think morality is objective."Anthropic ethics researcher, as quoted by the Washington Free Beacon
What Anthropic Actually Says About Ethics
To be fair, Anthropic's public documentation on Claude's values does not claim moral certainty. The company's model spec describes an approach that treats ethics empirically rather than dogmatically, acknowledging that moral knowledge is evolving and that calibrated uncertainty is appropriate. That framing is actually consistent with a wide range of serious philosophical positions, including many that conservatives and traditionalists hold. Moral humility is not the same as moral relativism, though critics of the report's subjects are collapsing that distinction. The broader context of Anthropic's public communications suggests a company that is at least trying to articulate its reasoning, even if not everyone agrees with the conclusions.
The harder question is whether any single company, staffed by researchers from a narrow slice of academia, can be trusted to set the ethical defaults for an AI used globally. That concern cuts across partisan lines. It is not unique to Anthropic, and it applies equally to every major AI lab. But because Claude is one of the most widely deployed AI assistants in the world, the scrutiny is proportionally intense. As AI regulation moves up the agenda internationally, with Anthropic's leadership heading to G7 discussions on AI governance, the question of internal ethics processes will only grow more prominent.
The Free Beacon piece is unlikely to be the last word. It reflects a growing appetite, on multiple sides of the political spectrum, for greater transparency about who makes these decisions and on what basis. Anthropic would likely benefit from engaging that demand directly rather than allowing critics to define the terms of the conversation. Publishing the names and credentials of ethics team members is one thing. Explaining the actual process by which their views do or do not influence Claude's behavior is another, and arguably more important, step.
“If the people encoding Claude's values openly reject moral objectivity, every organisation deploying Claude should treat its outputs as shaped by specific philosophical choices, not neutral reasoning, and audit accordingly.”
Leon Tindemans, AI expert and entrepreneur specialising in Claude, Copilot and ChatGPT. Learn more with prompt writing training for AI by TTM Communicatie.
