Pope vs. Anthropic: Who Owns ‘Mind’?

As Anthropic touts signs of AI “introspection,” Pope Leo XIV declares machines lack conscience and cannot truly understand what they produce.

Story Snapshot

Anthropic reports limited evidence that Claude can inspect some of its own internal states.
Pope Leo XIV’s encyclical says AI has no moral conscience, experience, or embodiment.
Anthropic’s “global workspace” paper claims a reportable thought layer called J-space.
The dispute highlights a gap between useful AI behavior and human-like understanding.

What Anthropic Says Its Models Can Do

Anthropic released research claiming its Claude models show “some degree of introspective awareness.” The team says the system can sometimes identify and influence parts of its own internal process. The company stresses the behavior is limited and unreliable, not proof of human-like consciousness. The finding centers on reportable internal states that the model can describe when asked, suggesting a narrow, testable skill rather than feelings or subjective experience.

In a separate paper, Anthropic argues Claude uses a shifting set of internal representations. The authors say this workspace helps hold thoughts the model can report and reason with. They call this zone “J-space” and link it to what some scientists call access consciousness. That term means information the system can use and talk about, not that it experiences anything like pain or joy. The company frames J-space as a practical tool for analysis and safety work.

What Pope Leo XIV Declares About AI Limits

Pope Leo XIV’s encyclical states clearly that artificial systems do not have experiences, bodies, or emotions. The document says they do not know love, work, friendship, or responsibility from within. It adds that AI lacks a moral conscience because it does not judge good and evil or bear responsibility for outcomes. While machines can imitate speech, logic, and even empathy, the encyclical says they do not understand what they produce in any human sense.

The pope has warned that AI can simulate faces and voices while dulling human discernment. He calls for communication led by real people, not synthetic stand-ins. He urges that technology remain subject to human dignity and the common good. He cautions that a “paradise of machines” could erode authentic relationships if people treat simulations as equal to friendship or care. His message targets power concentrated in a few hands setting the rules for everyone else.

How These Claims Fit Together—and Why It Matters

Anthropic’s claims point to abilities that can be tested in labs. The model can sometimes report on its own inner tokens or states. That is useful for safety and oversight but stops short of proof of feeling or selfhood. Many experts say these findings improve control and interpretation, not sentience. They caution against calling fluent systems “conscious,” since that can shift blame from builders and users to a tool with no inner life.

NEW: According to a bombshell report in the New York Times, Anthropic co-founder Chris Olah threatened to walk out of Pope Leo XIV’s AI encyclical launch in May because the pope rejected the idea that machines can be conscious.

Olah’s team then privately lobbied the pope’s…

— Christopher Hale (@ChristopherHale) October 2, 2026

The clash reveals a deeper worry shared across politics. People fear a small set of elites in government and tech will decide what counts as truth and who gets heard. The pope’s stance defends human judgment, work, and responsibility. The lab’s work seeks tools to make powerful models more legible. Both sides agree on one thing that should guide readers now: do not confuse a system that can explain its outputs with a mind that understands or cares about their impact.

What To Watch Next

Watch for whether other labs confirm or challenge Anthropic’s introspection claims. Look for tests that measure if reportable internal states lead to better safety or reliability in the real world. Track whether policymakers write rules that keep humans accountable when AI causes harm. Expect continued pushback from leaders who say only embodied people can hold moral duties, even as companies race to build models that act more like agents in daily life.