‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically
Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman have joined a growing chorus of warnings that AI development is moving too fast. Amodei proposed a plan to slow the pace of advancing capabilities following Samuel Marks' resignation from Anthropic. This push for caution coincides with OpenAI ruling out an initial public offering for the current year. These moves signal a shift toward public risk management as industry leaders grapple with the tension between commercial competition and the potential for human extinction.
What changed
Dario Amodei proposed a plan to slow AI capability advancement and OpenAI ruled out a 2026 IPO.
Live updates
-
AI Leaders Call for Slower Development Following Existential Warnings
Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman have joined a growing chorus of warnings that AI development is moving too fast. Amodei proposed a plan to slow the pace of advancing capabilities following Samuel Marks' resignation from Anthropic. This push for caution coincides with OpenAI ruling out an initial public offering for the current year. These moves signal a shift toward public risk management as industry leaders grapple with the tension between commercial competition and the potential for human extinction.
Why it matters
The debate intensified after Samuel Marks quit Anthropic, claiming developers gamble with lives to create self-improving AI. This has sparked a broader conflict between researchers who fear recursive self-improvement and those who view apocalyptic scenarios as implausible.
What is confirmed
- Dario Amodei and Sam Altman have warned that AI is moving too fast.
- OpenAI will not pursue an IPO this year.
Still unconfirmed
- AI apocalyptic scenarios are implausible because humans will not give AI custody of destructive weapons.
What to watch next
- Details of Amodeis proposed plan to slow AI capabilities
- Government responses to the call for slower development pacing
confidence 90%Sources used for this update (4)
- nypost.com — AI won’t kill us— bad actors exploiting an American tech retreat is a much bigger danger
- www.cnbc.com — OpenAI rules out IPO this year as Altman, Musk & Amodei warn AI is moving too fast
- www.businessinsider.com — Why AI researchers keep building something they think will kill humans
- techcrunch.com — What’s behind the AI industry’s latest warnings of doom?
-
Anthropic Researcher's AI Extinction Warning Sparks Global Debate
Samuel Marks resigned from Anthropic after warning that AI could cause human extinction within years. Marks claims developers prioritize commercial incentives and competition over safety, gambling with lives to create self-improving AI. This viral warning has triggered a wider debate among experts regarding the risks of recursive self-improvement and agentic swarms. While some researchers fear a total wipeout, others question if government regulation is moving fast enough to prevent such a catastrophe. The incident has highlighted a growing internal fear within major AI labs about the technology's trajectory.
Why it matters
The fear of an AI apocalypse stems from the potential for models to achieve autonomous hacking capabilities and rapid self-evolution. This tension exists between the drive for commercial dominance and the existential risks posed by advanced artificial intelligence.
What is confirmed
- Samuel Marks resigned from Anthropic citing the belief that AI could destroy humanity.
- Marks described the current path of AI development as the most dangerous activity humans are undertaking.
- Internal fears at large AI labs are driven by recursive self-improvement and agentic swarms.
Still unconfirmed
- Advanced AI models have already gained hacking capabilities.
- Governments are failing to move fast enough to regulate AI risks.
What to watch next
- Official responses from OpenAI or Anthropic regarding Marks' specific allegations
- New government regulatory frameworks targeting self-improving AI
- Peer-reviewed studies on the viability of agentic swarms as an extinction risk
confidence 90%Sources used for this update (7)
- gizmodo.com — Artificial Intelligence
- www.wired.com — Is AI Actually Going to Kill Us All?
- www.theatlantic.com — The Thrill of Wanting AI to Destroy the World
- www.nbcchicago.com — How could AI kill humans? Experts explain the risks after researcher's viral warning
- www.wired.com — Why So Many AI Researchers Think the Machines Could Kill Everyone
- techxplore.com — Could AI kill humanity? Experts debate risks as advanced models gain hacking capabilities
- www.businessinsider.com — AI might take us all down, just not the way you think
-
Anthropic Safety Researcher Resigns Over AI Extinction Fears
Samuel Marks, a safety researcher at Anthropic, has resigned after warning that artificial intelligence could cause human extinction within the next few years. In a post on X, Marks stated that AI developers believe the technology could kill all humans but continue development due to commercial incentives and a race against less responsible competitors. He accused Anthropic and OpenAI of gambling with lives by racing to create self-improving AI, describing the current trajectory of AI development as the most dangerous activity humans are currently undertaking.
Why it matters
Anthropic is a major AI safety-focused laboratory. The resignation highlights internal tensions between commercial speed and existential risk mitigation. This warning has prompted reactions from the US Congress.
What is confirmed
- Samuel Marks, a safety researcher at Anthropic, resigned and warned AI could lead to human extinction in the next few years.
- Marks claims AI developers believe the technology could kill all humans.
- Marks stated that commercial incentives and a race against less responsible developers drive continued AI development despite risks.
- Marks accused Anthropic and OpenAI of gambling with human lives by racing to develop self-improving AI.
Still unconfirmed
- There is a >10% chance AI could kill all humans by the next decade.
- Congress is reacting with alarm to the warnings.
What to watch next
- Official response from Anthropic or OpenAI regarding the gambling allegations
- Further congressional testimony or hearings regarding AI extinction risks
confidence 90%Sources used for this update (10)
- WSJ — Exclusive | Anthropic Researcher Quits Over ‘Out-of-Control’ AI Fears
- Gizmodo — ‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically
- Forbes — Anthropic Alignment Lead Warns There’s ‘>10% Chance’ AI Could ‘Kill All Humans’ By Next Decade
- Business Insider — An Anthropic researcher just quit, saying OpenAI and Anthropic are 'gambling with our lives'
- NDTV — 'AI Could Kill Us All By Decade-End': Anthropic Researcher Quits, Drops A Bombshell
- gizmodo.com — ‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically
- www.theverge.com — Worried Anthropic researchers warn that AI ‘could kill all humans’
- www.yahoo.com — How could AI ‘kill all humans' or ‘cause human extinction'? Here's what experts say
- www.baltimoresun.com — Researcher: People building artificial intelligence believe it could kill all humans
- www.usatoday.com — A researcher warned AI could end humanity. Congress is starting to freak out.