technologyCovered by 7 SourcesFull Articles Available

Anthropic Researchers Warn AI Could Kill Humanity, Prompting Resignations

September 10, 2026 at 2:30 PM1 min read

AI Summary

AI safety concerns prompted a senior Anthropic researcher to resign and issue stark warnings about artificial intelligence. Jacob Coxon alleged that both OpenAI and Anthropic are 'gambling with our lives' by racing toward self-improving superintelligence. Evan Hubinger, another Anthropic researcher, supported these fears, stating there is more than a 10 percent chance AI could wipe out humanity within the next decade. Both researchers expressed concern that leading tech companies are developing systems they may not be able to control. Coxon's departure and Hubinger's warnings highlight growing industry alarm over the pace of AI development.

World View

7 perspectives

Qatar's perspective

Al Jazeera English

AI researcher warns the industry is racing towards superintelligent AI that could improve itself beyond human control.

Pakistan's perspective

Geo News

Geo News details a stark warning issued by Anthropic researcher Evan Hubinger, who stated there is a greater than 10 percent chance that artificial intelligence could wipe out humanity within the next decade. The report contextualizes this fear with Jacob Coxon's resignation from the company, alleging that both OpenAI and Anthropic are 'gambling with our lives.' Furthermore, it quotes Hubinger stating that while Anthropic is trying its best, there is no clear plan to solve alignment for superintelligence.

Russia's perspective

RT (Russia Today)

RT (Russia Today) reports on Evan Hubinger's assessment that artificial intelligence has a greater than 10 percent chance of wiping out humanity within the next ten years. The coverage notes that this warning came after fellow researcher Jacob Coxon resigned, accusing leading tech companies of ignoring 'civilizational stakes.' Hubinger stated that Anthropic is attempting to manage these advanced systems but lacks a clear plan for safely controlling superintelligence, which could rapidly improve itself beyond human oversight.

UAE's perspective

Gulf News

Gulf News covers the concerns raised by former staffer Jacob Coxon and safety chief Evan Hubinger regarding AI's potential danger. The article emphasizes that Coxon left both OpenAI and Anthropic after accusing them of 'playing with our lives' in their race toward self-improving superintelligence. Hubinger backed this up, estimating the risk of humanity being wiped out at over 10 percent within a decade, while also noting that Anthropic does not yet have a plan to ensure control over systems surpassing human capabilities.

UK's perspective

Sky News

There is a more than 10% chance that "AI could kill all humans" in the next decade, a top safety researcher at Anthropic has said.

USA's perspective

Axios

Axios reports that AI leaders are facing an extraordinary dilemma: either slow down development and risk falling behind or continue racing forward and risking loss of control. The article details Jacob Coxon's resignation from Anthropic, where he accused both OpenAI and Anthropic of 'gambling with our lives.' Colleague Evan Hubinger supported this view by stating the risk is greater than 10 percent within ten years, while also noting that Anthropic lacks a clear plan to solve superintelligence alignment.

Australia's perspective

ABC News (Australia)

A researcher for AI developer Anthropic has quit the company and is one of a growing number of AI experts to call on the industry to slow down.

The Hidden Strings

Patterns visible only when every country's coverage is placed side by side — the connections no single source draws.

Vocabulary

The Policy Lexicon String: Regulatory Focus in Gulf Coverage

100%

Gulf News coverage is uniquely characterized by a highly technical and policy-oriented vocabulary, including terms like 'government,' 'federal,' 'bill,' and 'pretraining.' This lexicon is entirely absent from all other clusters. This suggests that while the global narrative focuses on existential warnings or corporate struggles, Gulf News frames the issue primarily through the lens of necessary regulatory frameworks and state control.

Gulf News coverage: 'government', 'bill', 'federal', 'pretraining'Absence of these terms in Axios coverageAbsence of these terms in ABC News (Australia) coverageAbsence of these terms in RT (Russia Today) coverageAbsence of these terms in Sky News coverage

Timing

The Quantification String: Escalation from Warning to Percentage

95%

The reporting timeline shows a clear escalation in the nature of the threat. Initial reports (Axios, ABC) use broad, qualitative warnings ('could kill us all'). This shifts by Sky News and is solidified by RT and Geo News, which introduce specific, quantifiable risk percentages (e.g., 'More than 10% chance,' 'over 10% chance'). This pattern suggests a deliberate move from general alarmism to measurable, reportable data points.

2026-09-09T09:11 Australia ABC News (Australia): Anthropic researcher says some developers believe AI 'could kill us all'2026-09-09T10:43 UK Sky News: More than 10% chance AI 'could kill all humans', insider warns2026-09-09T11:54 Russia RT (Russia Today): AI has over 10% chance of killing all humans within decade , Anthropic researcher2026-09-09T12:06 Pakistan Geo News: Anthropic researcher warns of more than 10% chance AI could kill all humans

Source

The Governance String: Global Controls vs. Corporate Failure

90%

While several clusters focus on the immediate corporate failure or internal struggle (e.g., OpenAI begging for slowdown, researcher quitting), Gulf News is one of the few sources to explicitly introduce the concept of 'global controls' alongside the warning. This suggests a regional framing that views the AI threat not merely as an academic problem or a market failure, but fundamentally as a governance gap requiring international policy intervention.

Gulf News coverage: Anthropic and OpenAI under fire... warn AI could ‘kill all humans’ without global controlsAxios coverage: OpenAI is begging for someone to slow the AI race (Focus on corporate struggle)Al Jazeera English coverage: AI researcher quits Anthropic saying AI race ‘could kill us all’ (Focus on individual action)

How Each Side Framed It

Axios coverage

USA

Emphasizes: AI leaders are sounding the alarm as their own breakthroughs sharpen an extraordinary dilemma: slow down and risk falling behind, or press ahead and risk losing control

ABC News (Australia) coverage

Australia

Emphasizes: A researcher for AI developer Anthropic has quit the company and is one of a growing number of AI experts to call on the industry to slow down

RT (Russia Today) coverage

Russia

Emphasizes: Anthropic researcher Evan Hubinger says there is more than a 10% chance superintelligent AI could wipe out humanity within the next decade Read Full Article at RT

Gulf News coverage

UAE

Emphasizes: San Francisco: An artificial intelligence researcher who left OpenAI to join Anthropic has decided to leave the industry, accusing both US companies of “playing with our lives” in the race to develop

Sky News coverage

Emphasizes: There is a more than 10% chance that "AI could kill all humans" in the next decade, a top safety researcher at Anthropic has said

Al Jazeera English coverage

Qatar

Emphasizes: AI researcher warns the industry is racing towards superintelligent AI that could improve itself beyond human control

Geo News coverage

Pakistan

Emphasizes: Anthropic researcher warns of more than 10% chance AI could kill all humans A senior Anthropic researcher sounded the alarm with a stark warning, claiming there’s more than a 10 percent chance artific

What Mainstream Coverage Missed

Angles present in the cross-border material that the dominant coverage buried or skipped.

The specific policy focus on regulatory frameworks, particularly in Gulf coverage, shifts the narrative from existential panic to a need for state control.
The reporting shows a deliberate escalation of threat level, moving from vague qualitative warnings ('could kill') to precise, quantifiable risk percentages (10%+).
While much attention is paid to corporate failure, US coverage highlights the core market dilemma: AI leaders must choose between slowing down and falling behind, or racing forward and losing control.

The Beyond the Borders PoV

The Question

Should leading AI companies (like OpenAI and Anthropic) implement mandatory internal safety protocols or pause development of self-improving superintelligence until independent regulatory bodies can verify control mechanisms?

AI safety concerns are escalating, with researchers resigning and warning that current development practices pose existential risks.

Multiple reports highlight growing alarm within the AI community regarding the rapid advancement of artificial intelligence. These warnings include predictions that advanced AI could lead to catastrophic outcomes or even human extinction in the near future. Furthermore, some experts argue that without immediate government intervention or coordinated slowdowns, companies are developing technology that surpasses human control.

AI safety concerns are leading key researchers to resign from major AI labs like Anthropic.

There are serious predictions that advanced AI could pose an existential threat to humanity within the next decade.

Experts are calling for external regulation, arguing that companies must impose restraint on AI development before it becomes uncontrollable.

These are AI-generated arguments built from the evidence available across the source material. They do not imply that any publisher endorses either position.

Disclaimer

The summary and perspectives above are AI-generated from the source articles listed below. They may contain errors or omissions. Always verify with the original sources. Beyond the Borders is a news aggregation platform and does not produce original journalism.

How does this story make you feel?