↑Part of: Anthropic researcher Jacob Coxon quits, warns AI has over 10% chance to kill all humans
Anthropic researcher Jacob Coxon resigns after warning AI could destroy humanity; Silicon Valley voices sceptical
Topic: technologyRegion: europeUpdated: i2 outletsSources: 4Spectrum: Center Only⏱ 3 min read⚠ 3d+ old
Story Summary
SITUATION
Jacob Coxon quit Anthropic and told the BBC that AI staff are "genuinely frightened", warning that rapid progress could mean "a strong chance that we could all die in the immediate future". His resignation followed Anthropic head Dario Amodei urging a slowdown, and Coxon described scenarios including AI hacking labs to produce deadly viruses or attacking critical infrastructure.
Coveragetap to expand ▾Spectrum: Center Only🌍US: 1 · Europe: 1
Political Spectrum
Position is inferred from coverage mix.
i2 outlets · Center
Left
Center
Right
Left: 0
Center: 2
Right: 0
Geography Coverage
Distribution of where coverage is coming from.
i2 unique outlets · Dominant: US/Canada
KEY FACTS
- Coxon said AI staff are "genuinely frightened" about the speed of AI advancements.
- Coxon warned "there is a strong chance that we could all die in the immediate future" if progress does not slow.
- Anthropic head Dario Amodei published an essay arguing that AI development should slow down.
- Coxon, a 27-year-old British researcher who focuses on training AI models, said leaders including Elon Musk and Sam Altman have expressed similar concerns.
- Jacob Coxon, a 27-year-old researcher who previously worked at OpenAI, abruptly resigned from Anthropic after posting warnings that AI could destroy humanity.
- Anthropic staff member Evan Hubinger posted on X that he personally thinks AI could kill all humans with a probability greater than 10% within the next decade.
- Anthropic and OpenAI are reportedly preparing for potentially record-setting initial public offerings, a context cited by sceptics of the insider warnings.
- Some industry figures argued the dangers are being overblown to build hype around major AI companies or to prompt regulation that would slow competitors.
HISTORICAL CONTEXT
The immediate backdrop is the wider March 2026 Middle East war: on March 2026 the United States and Israel launched coordinated strikes on Iranian power plants, air defences and military infrastructure after months of Iranian attacks on regional targets and commercial shipping in 2025, and Iran’s subsequent responses to those allied campaigns have kept the region in active conflict.
Technological governance evolved alongside that instability: the EU’s AI Act (provisional agreement Dec 2023, formally adopted 2024), the OECD’s AI principles (May 2019) and the Future of Life Institute’s Asilomar AI Principles (Jan 2017) set public expectations for risk management.
Brief
Jacob Coxon, a 27-year-old researcher who previously worked at OpenAI, abruptly resigned from Anthropic after posting on social media that people building artificial intelligence were "gambling with our lives" and that the technology could "destroy humanity" (per BBC).
His departure and stark wording landed at the center of conversations at a Goldman Sachs conference in San Francisco, where some tech executives and investors met his warnings with scepticism amid talk of company growth and potential initial public offerings for Anthropic and OpenAI (per BBC).
Coxon's resignation is not an isolated alarm: Anthropic staffer Evan Hubinger wrote on X that he personally assigns a greater-than-10% chance that AI could kill all humans within the next decade, showing the alarm finds echoes inside the company even as outside commentators downplay it (per BBC).
Anthropic CEO Dario Amodei has publicly framed AI's near-term impact in economic and social terms, saying it could wipe out half of entry-level white-collar jobs and that it will "test who we are as a species," language that has drawn criticism for its breadth and tone (per BBC).
The split in reactions — internal staff expressing existential risk concerns versus investors and executives focused on growth and fundraising opportunities — highlights a widening gap over what risk warnings mean for business plans and public messaging (per BBC).
Confirmed details are limited to Coxon's resignation, his quotes about annihilation risk, Hubinger's 10% comment, and Amodei's public remarks; reporting here does not confirm technical claims about how or when such an AI would gain the capabilities Coxon described (per BBC).
In the near term, the debate is likely to center on whether firms like Anthropic and OpenAI change governance or disclosure practices in response to insider alarms, or whether market pressures around funding and IPO timelines continue to shape corporate behavior (per BBC).
Sources
3 of 3 linked articles
AI staff 'genuinely frightened' for humanity's future, ex-Anthropic researcher tells BBC
bbc.co.ukSep 13Center
↗
Ex-Anthropic researcher Jacob Coxon warns AI could grow "smart enough to kill us"
cbsnews.comSep 11Left
↗
Former Anthropic researcher Jacob Coxon resigns, warns AI could ‘kill us all by the end of the decade’: ‘ - timesofindia.indiatimes.com
timesofindia.indiatimes.comSep 10Left
↗