OpenAI CEO Sam Altman has made clear he disagrees with Anthropic's decision to consult religious scholars as part of its effort to define Claude's values and identity. The criticism, which surfaced on social media, adds a new layer to the ongoing rivalry between the two prominent AI laboratories and raises pointed questions about how AI companies should approach questions of machine ethics and character.
What Anthropic Actually Did
Anthropic flew in a group of religious thinkers and ethicists to its offices to participate in discussions about the moral framework underpinning Claude. As we previously reported, the company framed this as part of a broader effort to incorporate diverse perspectives into how the model reasons about sensitive topics, values, and its own sense of purpose. The sessions were not intended to make Claude religious, but rather to stress-test its ethical reasoning through the lens of traditions that have grappled with questions of moral identity for centuries.
Key Facts
- Anthropic brought religious scholars to its offices to consult on Claude's values and identity.
- OpenAI CEO Sam Altman publicly criticized the initiative.
- Anthropic has previously stated it wants Claude to have a stable, grounded character.
- The dispute highlights differing philosophies between the two companies on AI alignment.
- Neither Anthropic nor OpenAI has released a formal statement addressing the disagreement directly.
Altman's objection, while brief in public, signals a real philosophical gap. OpenAI has generally taken a more technically focused approach to alignment, relying on reinforcement learning from human feedback and constitutional methods. Bringing in theologians and religious ethicists is a departure from that paradigm, and Altman appears to find the move unnecessary or misguided. His comments echo past moments where he has not hesitated to weigh in on competitor decisions, including when he mocked an Anthropic advertisement that drew widespread criticism for its tone.
Anthropic has said it wants Claude to have a settled sense of identity, one that does not crumble under pressure or manipulation. Consulting traditions with long histories of thinking about selfhood and moral duty is, from that angle, a logical step.ClaudeAINews.com analysis
A Deeper Disagreement About AI Character
The argument is not simply about religion. It cuts to a more fundamental question: should AI systems be designed with something resembling a coherent inner life, shaped by humanistic and philosophical traditions, or should alignment remain a purely technical exercise? Anthropic has consistently argued for the former. The company's model documentation has described Claude as having genuine character traits, intellectual curiosity, and a stable ethical orientation that persists across interactions.
Critics of Anthropic's approach, and Altman appears to be among them, may argue that framing AI development in quasi-spiritual terms risks obscuring the engineering realities of how these systems actually work. There is also a concern that importing religious frameworks could introduce biases or value systems that are not universally shared. The debate is unlikely to be resolved quickly, and it is playing out against a backdrop of increasing scrutiny over how AI labs make decisions about model behavior and values. Questions about those decisions are increasingly reaching policymakers, as seen when executives from both companies were called to address world leaders at forums like the G7 summit in France.
For now, Anthropic has not backed away from its position. The company continues to invest in what it considers a principled, humanities-informed approach to building Claude's character. Whether that approach produces a meaningfully better or safer AI remains an open question, but it is one the broader industry is watching closely. The exchange between Altman and Anthropic's implicit response through its actions is the kind of public disagreement that rarely happened when these companies were more careful to avoid direct confrontation.
