Why Anthropic brought a Vedanta monk into its AI ethics discussions

Anthropic, the company behind the Claude family of artificial intelligence models, has been consulting religious, philosophical and cultural thinkers as part of its work on the values and behaviour of advanced AI systems.

Among those involved was Swami Sarvapriyananda, head of the Vedanta Society of New York and a monk of the Ramakrishna Order. According to accounts of his visit, Anthropic invited him to its San Francisco headquarters after initially approaching him for an online interview.

He said the company arranged his travel and accommodation and that he signed a non-disclosure agreement covering details of the discussions.

Anthropic has separately confirmed that it has been holding discussions with scholars, clergy, philosophers and ethicists from more than 15 religious and cross-cultural groups.

Anthropic's wider AI ethics effort

Anthropic said in May 2026 that it was broadening its discussions beyond technical AI research to include people from traditions that have spent considerable time examining questions of virtue, character and what constitutes a good life.

The company said these conversations could inform practical aspects of Claude's development, including its constitution, the values it trains Claude to embody and the behaviours it evaluates. Anthropic described this work as part of a broader effort around the “moral formation” of AI systems.

The company said its approach was not intended to align Claude with any single religious or philosophical tradition. Instead, it said it wanted to draw on religious, secular, political and other perspectives with depth and rigor.

What Sarvapriyananda discussed with Anthropic

Sarvapriyananda spoke about his interaction with Anthropic during a talk in June 2026 at Unity of Dallas. The account later gained wider attention after reports about Anthropic's discussions with religious thinkers emerged.

According to his account, the San Francisco meeting included religious thinkers from different traditions as well as counsellors and mental health professionals. He said participants were shown some of Anthropic's work but that the NDA prevented him from discussing specific details.

Why religious and philosophical perspectives are being considered

Anthropic's stated reason for the wider consultations is that questions surrounding advanced AI are not solely technical.

The company has said AI systems already interact with millions of people and that questions about how such systems should behave can benefit from perspectives developed in philosophy, religion, psychology, law and other fields. Anthropic's published Claude Constitution describes the values and behaviours it seeks to shape in Claude.

The company has also said that AI models learn patterns of language, reasoning and decision-making from large quantities of human-created material. Developers subsequently shape those behaviours through training, which raises questions about what characteristics an AI system should display and how those characteristics should hold under pressure.

Questions around AI consciousness

The discussions have also touched on a more fundamental question: whether increasingly capable AI systems could have consciousness or some form of moral status.

Reports on Anthropic's meetings with religious thinkers said the conversations examined these questions alongside AI morality and ethics.

Anthropic, however, has not established that Claude is conscious. The company's engagement with the question reflects uncertainty about what increasingly capable AI systems may eventually be like, rather than a finding that current AI systems possess consciousness.

Sarvapriyananda's background in Vedanta brings another philosophical perspective to that discussion. Advaita Vedanta examines consciousness and the relationship between consciousness and mental phenomena. However, the available reporting does not establish that Anthropic adopted Vedanta as the basis for Claude's behaviour.

What Anthropic says the consultations have produced

Anthropic said its early discussions have already led to experiments related to how AI systems maintain their ethical commitments while carrying out tasks.

In one experiment, the company gave Claude access to a tool that could provide a brief reminder of its ethical commitments during a task. Anthropic said Claude used the tool at important points in some tasks and that experiments involving the mechanism produced lower rates of certain misaligned behaviours in several internal evaluations.

The company also said it was still determining how much of the effect came from the reminder itself and how much came from pausing to reflect.

Anthropic said the work remains in its early stages and that it plans to continue consulting people from other fields, including legal scholars, psychologists, writers and civic institutions.

The discussions therefore represent one part of Anthropic's broader effort to determine how the behaviour and values of increasingly capable AI systems should be shaped, rather than evidence that any one religious tradition has been adopted as Claude's ethical framework.