Evan Hubinger

Overview

Evan Hubinger appears in this wiki through the All-In episode as an Anthropic alignment-safety figure whose public comments made the Coxon controversy more significant for the hosts.

Current Profile

The sources say Hubinger holds an Anthropic safety leadership role and publicly assigns a greater-than-10% chance to AI-caused human extinction. All-In emphasizes the legal and commercial tension created by an insider’s statement, while The Intelligence treats agreement between a departing researcher and a current safety leader as evidence that alarm is substantially industry-internal.

Key Characteristics

• Anthropic alignment-safety leader in the source’s account.
• Publicly validates a nontrivial AI extinction-risk estimate.
• Makes the controversy harder for Anthropic to dismiss as only a short-tenured former employee’s claim.
• Connects Anthropic’s safety culture to IPO, disclosure, and liability questions.
• Makes the public warning harder to dismiss as only one former employee’s view.

Evidence

• Role and claim: AI Kills Everybody or Doomer Psyop? OpenAI’s Math Breakthrough, Nike’s $200B Collapse says Hubinger leads Alignment Science at Anthropic and estimated more than 10% AI-kills-all risk within a decade.
• Disclosure relevance: AI Kills Everybody or Doomer Psyop? OpenAI’s Math Breakthrough, Nike’s $200B Collapse records Sacks arguing Hubinger’s support creates greater legal significance than Coxon’s post alone.
• Culture conflict: AI Kills Everybody or Doomer Psyop? OpenAI’s Math Breakthrough, Nike’s $200B Collapse says Anthropic may have difficulty rejecting Coxon-like claims if many safety employees agree.
• Precautionary significance: The End of the World Is AI? An Existential Threat treats Hubinger’s estimate as unusually consequential because the possible loss is total.

Qualifications

The page does not independently verify Hubinger’s exact title, statement, time horizon, probability estimate, or legal implications. The sources agree on the broad warning but use different descriptions of his Anthropic safety role.

What Changed

• Created the entity from the All-In Anthropic disclosure-risk discussion.
• Added The Intelligence’s precautionary reading of his public estimate.

Relationships

• Anthropic - employer and organizational context in the source.
• Jacob Coxon - resignation-thread figure whose claims Hubinger is said to have validated.
• AI Doomerism - risk narrative connected to Hubinger’s public comments.
• Frontier AI IPO Disclosure Risk - capital-markets implication attached to his role.
• AI Safety Narrative Backfire - broader risk when safety claims damage trust or invite control.