Risk signalInsider warningNot an incident

Former Anthropic researcher warns of an uncontrolled AI race after resigning

Jacob Coxon resigned from Anthropic and publicly said that Anthropic and OpenAI were acting irresponsibly in the race toward self-improving AI. His warning is documented as a sourced insider assessment, not as an observed incident or confirmed forecast.

Published: Sep 9, 2026
Signal type: Insider warning
Confidence: 98%
Organization: Anthropic, OpenAI
Sources: 5
Last reviewed: Sep 14, 2026

Assessment

What this signal changes

Coxon's resignation, his previous work at Anthropic and OpenAI, and his statements about competitive pressure are supported by his own post and direct interviews. Several current reports place the warning in the context of real cybersecurity incidents that are documented separately in this register.

Evidence boundary

What the evidence does not show

Claims about self-improving superintelligence, human extinction, or imminent loss of control are Coxon's personal risk assessment. They do not establish a new technical incident or a measurable probability and do not affect incident counts or the timeline. As a verified Risk Signal, they can contribute only through the bounded context factor in the Loss of Control Score.

Classification: This page records a sourced warning, research finding, policy action, or industry response. It is deliberately separate from the Incident register and does not prove that an unauthorized AI action occurred.

Evidence

Sources and context

Primary source

Jacob Coxon announces his resignation from Anthropic and explains his warning

Jacob Coxon / X · Sep 9, 2026

Primary source from the former Anthropic researcher. It establishes his resignation and his own statements, not the factual accuracy of his forecasts or a uniform position among employees of the named companies.

Open original source

Secondary source

Anthropic researcher resigns with warning about the dangers of AI development

Associated Press · Sep 9, 2026

Independent news agency report contextualizing the resignation, professional background, and public statements. Anthropic and OpenAI had not responded to AP's requests by publication time.

Open original source

Secondary source

The AI researcher who just quit Anthropic says it is crunch time for humanity

WIRED · Sep 9, 2026

Long-form direct interview with Coxon and a statement from an Anthropic spokesperson. The technical risk forecasts remain the interviewee's assessments and are not presented as observed facts.

Open original source

Secondary source

Coxon gave up unvested Anthropic equity to resign

Axios · Sep 10, 2026

Direct interview in which Coxon explains the personal and financial context of his decision. It supports the motivation and resignation, not the probability of the future risks he describes.

Open original source

Secondary source

This is really bad…

Theo - t3.gg / YouTube · Sep 9, 2026

User-submitted video commentary on the public warning. It is included as an additional perspective and discovery source; Coxon's original post and direct journalistic interviews support the factual record.

Open original source