Anthropic Co-Founder Says AI Systems Show 'Introspection' And Emotion-Like States At Vatican Event, Prompting Vinod Khosla To Blast 'Elitist' Debate

On Monday, Anthropic co-founder Chris Olah said that advanced AI systems displaying patterns resembling "joy," "fear" and "grief," after which venture capitalist Vinod Khosla criticized the Vatican-led discussion as "elitist."

AI Ethics Debate Takes Center Stage At Vatican

Pope Leo XIV has released a major encyclical titled "Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence," warning that AI development must slow down and remain under human control.

The pope cautioned against concentrating AI power "in the hands of only a few people" and argued that "merely regulating it is insufficient."

He also called for "disarming AI," saying the goal was not to reject technology but to prevent it from "dominating humanity."

The Vatican unveiled the document alongside Olah, who spoke about the growing complexity and unpredictability of advanced AI systems.

Chris Olah Warns Of ‘Mysterious' AI Behavior

During the presentation, Olah said researchers studying the internal mechanics of AI models continue to uncover behaviors they do not fully understand.

"We keep finding things that are mysterious, even unsettling," Olah said, adding that researchers have observed structures resembling patterns seen in human neuroscience as well as evidence of "introspection."

He further said AI systems appear to develop internal states that "functionally mirror joy, satisfaction, fear, grief, and unease," though he acknowledged researchers still do not know what those findings ultimately mean.

Olah urged governments, religious groups and civil society to become more involved in AI oversight, saying the industry needs "moral voices that the incentives cannot bend."

Pope Leo XIV thanked Olah and said the Church was ready to work alongside AI researchers to help guide humanity through the rise of AI.

Vinod Khosla Calls Debate ‘Elitist'

Khosla criticized the framing of the Vatican discussion in a post on X, calling it "very elitist."

Khosla argued that questions about how AI should interact with society should be answered broadly by humanity rather than by "religions," "philosophy," or "elitist institutions with their view of superiority."

Vance Warns Of AI Cyberattack Risks

Vice President JD Vance has cautioned executives from Microsoft Corp (NASDAQ:MSFT), Alphabet Inc. (NASDAQ:GOOG) (NASDAQ:GOOGL), OpenAI and Anthropic that Mythos— an AI system capable of independently identifying software vulnerabilities — could potentially enable cyberattacks targeting community banks, hospitals and water treatment facilities.

At last year's Paris AI summit, Vance had warned that excessive regulation could stifle the AI industry.

Earlier this month, National Economic Council Director Kevin Hassett said that Donald Trump is weighing an executive order that would introduce oversight measures for new artificial intelligence models.

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Photo Courtesy: Photo For Everything on Shutterstock.com