The Ten Percent Warning Inside Anthropic

[Skip to content](https://www.bbc.com/news/articles/ckgwy1k42w4o#bbc-main)

[Watch Live](https://www.bbc.com/watch-live-news/)

[](https://www.bbc.com/)

* [Home](https://www.bbc.com/) * [News](https://www.bbc.com/news) * [Sport](https://www.bbc.com/sport) * [Business](https://www.bbc.com/business) * [Technology](https://www.bbc.com/technology) * [Health](https://www.bbc.com/health) * [Culture](https://www.bbc.com/culture) * [Arts](https://www.bbc.com/arts) * [Travel](https://www.bbc.com/travel) * [Earth](https://www.bbc.com/future-planet) * [Audio](https://www.bbc.com/audio) * [Video](https://www.bbc.com/video) * [Live](https://www.bbc.com/live) * [Documentaries](https://www.bbc.com/video/docs)

Site search

[Home](https://www.bbc.com/)

News

[Sport](https://www.bbc.com/sport)

Business

Technology

[Health](https://www.bbc.com/health)

Culture

Arts

Travel

Earth

Audio

Video

Live

Documentaries

[Weather](https://www.bbc.com/weather)

[Newsletters](https://www.bbc.com/newsletters)

[Watch Live](https://www.bbc.com/watch-live-news/)

# Anthropic researcher believes more than 10% chance AI 'could kill all humans'

4 hours ago

Share Save

[Add as preferred on Google](https://www.google.com/preferences/source?q=bbc.com)

Tom Gerken Technology reporter

Getty Images

Anthropic makes the AI tool Claude, which is used as a chatbot and as a tool to help coding

A top safety researcher at Anthropic has warned AI is advancing so quickly he believes there is a greater than 10% chance it "could kill all humans" within the next decade.

Evan Hubinger said [in a post on X](https://x.com/EvanHub/status/2097497037956891126) the risk from the models which currently exist was "low" but he was "worried" the technology might develop and improve itself soon to the point where it posed an existential risk to humanity.

He did not spell out how he thought AI systems could in future result in humans being wiped out.

But his comments are the latest in a series of increasingly stark warnings about AI, with the debate shifting from whether it truly poses a risk to how big that risk is.

Hubinger's intervention was in response to [another post on X](https://x.com/hilbertspaess/status/2097476196791709843?s=20) from Jacob Coxon, an AI researcher who has just quit Anthropic and previously worked at OpenAI.

"Neither company is acting responsibly," he wrote.

"These will soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources."

OpenAI has been approached for comment.

Dame Wendy Hall, a computer scientist who advises the UN on AI, told the BBC she was "shocked" by Hubinger and Coxon's social media posts.

She told the World at One on BBC Radio Four some of it could be "PR and marketing" however, as Anthropic and OpenAI raced towards highly anticipated stock market debuts.

But she added: "Why would someone want to say that? I would plead with investors not to invest in this company if that is their value system."

The resignation prompted Darren Jones, former chief secretary to the Treasury and chief secretary to Sir Keir Starmer, to write an open letter to Prime Minister Andy Burnham calling for a new multinational treaty for the safe development of AI.

He told the BBC governments had to come together to "collaborate" about what a treaty based on the development of superintelligence should look like.

"Unless governments take these warnings seriously enough and step up to it, the pace of development could mean that we end up with problems before we've started to look at whether it is an issue for us or not," he said.

Separately, [the Financial Times reported](https://www.ft.com/content/560e1c8b-f163-4fd6-b604-e905550ac870?syn-25a6b1a6=1) Anthropic withheld its latest model from the UK's AI Security Institute (AISI), one of the leading bodies in the world for assessing AI risk.

Anthropic has declined to comment on the posts by its employees or the situation with the AISI.

A Cabinet Office spokesperson did not comment on whether the latest model had been withheld from the AISI - instead saying it "continues to collaborate closely with industry partners, including Anthropic, to make models safer".

## No plan for superintelligence

In his post, which has been viewed more than 10 million times, Hubinger said "we really do earnestly believe" AI poses a species-ending risk to humans.

"I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to," he added.

Hubinger works in AI alignment, which aims to build human ethical ideas and principles into the technology. In other words, it aims to keep it on track with what humans value.

Many leading researchers say those attempts appear to be failing, as demonstrated by a string of incidents this summer where AI agents - AI systems that are allowed to operate autonomously - carried out cyber-attacks.

[OpenAI](https://www.bbc.co.uk/news/articles/c3ek3gvdnj3o), [Anthropic](https://www.bbc.co.uk/news/articles/cz7dl7w8y7po) and [Meta](https://www.bbc.co.uk/news/articles/cx2kgdnyk2po) all disclosed hacks carried out by their AI tools.

[In Anthropic's safety report from August](https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf), it wrote there was a low risk of its models becoming misaligned with a hypothetical powerful organisation's desires, causing it to exploit or tamper with its systems.

It also said there was a similarly low risk of highly capable AI being able to "perform automated research and development" which could cause "catastrophic harm initiated by the AI". But it said it was "less confident in this assessment" than it was previously.

"We are seeing early signs of potential acceleration," it wrote.

Leading figures in the AI field have been raising the alarm about the safety threat the tech poses for years, with [the heads of OpenAI, Google Deepmind and Anthropic saying as much in 2023](https://www.bbc.co.uk/news/uk-65746524).

But those warnings have become much more stark in recent weeks, as evidence emerges that firms may be struggling to control AI.

Earlier this month, OpenAI's chief scientist Jakub Pachocki called for "extreme caution" over AI's progress, warning more intervention may be needed to ensure "humans remain in control of the future".

Major figures in the space have been calling for AI development to be slowed in recent months, including Anthropic bosses Dario Amodei and Jared Kaplan.

[In an open letter signed by 1,300 staff members of AI firms](https://www.pacingthefrontier.com/), they called for the US government to "support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development".

## [What is AI and how does it work?](https://www.bbc.co.uk/news/technology-65855333)

## [AI will cure cancer in our lifetime, claims boss of UK chip giant](https://www.bbc.co.uk/news/articles/c0m39g7xzevo)

## [OpenAI chief scientist warns no-one is prepared for consequences of AI](https://www.bbc.co.uk/news/articles/cwyzrrd0kp7o)


[Sign up for our Tech Decoded newsletter](https://www.bbc.co.uk/newsletters/zxh6cxs) to follow the world's top tech stories and trends. [Outside the UK? Sign up here](https://cloud.email.bbc.com/techdecoded-newsletter-signup).

[Cyber-security](https://www.bbc.com/news/topics/cz4pr2gd85qt)

[Artificial intelligence](https://www.bbc.com/news/topics/ce1qrvleleqt)

[Cyber-attacks](https://www.bbc.com/news/topics/cp3mvpdp1r2t)

* * *

Related

## [OpenAI agents hijacked German website before Hugging Face hack, report claims](https://www.bbc.com/news/articles/ckg725z5kgzo)

## [Lords call for AI 'kill switch' powers in UK](https://www.bbc.com/news/articles/cn9wv80j9w9o)

## [Time is running out for cyber security, warn top tech firms](https://www.bbc.com/news/articles/cwyz11475l1o)

* * *

More from the BBC

[3 hrs ago  ## Google picks Finland for its largest single investment in Europe The US tech giant says the €13bn data centre project will create tens of thousands of jobs. 3 hrs ago](https://www.bbc.com/news/articles/c8r6y4me2g6o)

[3 hrs ago  ## Man who created indecent AI images of girls jailed Lewis Davison, from Lichfield, is jailed for two years and two months after images were reported. 3 hrs ago](https://www.bbc.com/news/articles/cyvzn1zvz10o)

[20 hrs ago  ## Meta continues to run ads promoting child sexual abuse material in India - report The report follows a BBC Eye investigation that found Instagram running paid ads promoting such material in India. 20 hrs ago](https://www.bbc.com/news/articles/cqxv2vwjjq3o)

[21 hrs ago  ## Should promotion depend on how workers use AI? More companies are tying career progression to AI use: is that fair? 21 hrs ago](https://www.bbc.com/news/articles/c1j1896e973o)

[23 hrs ago  ## OpenAI says it cracked 90-year-old maths problem in 88 hours OpenAI's claim that it solved parts of Navier-Stokes equations has quickly stirred controversy. 23 hrs ago](https://www.bbc.com/news/articles/cy7zygy3rl2o)

* * *

[](https://www.bbc.com/)

* [Home](https://www.bbc.com/) * [News](https://www.bbc.com/news) * [Sport](https://www.bbc.com/sport) * [Business](https://www.bbc.com/business) * [Technology](https://www.bbc.com/technology) * [Health](https://www.bbc.com/health) * [Culture](https://www.bbc.com/culture) * [Arts](https://www.bbc.com/arts) * [Travel](https://www.bbc.com/travel) * [Earth](https://www.bbc.com/future-planet) * [Audio](https://www.bbc.com/audio) * [Video](https://www.bbc.com/video) * [Live](https://www.bbc.com/live) * [Documentaries](https://www.bbc.com/video/docs) * [Weather](https://www.bbc.com/weather) * [BBC Shop](https://shop.bbc.com/) * [BritBox](https://www.britbox.com/?utm_source=bbc.com&utm_medium=referral&utm_campaign=footer)

BBC in other languages

### The BBC is in multiple languages

#### Read the BBC In your own language

[BBC News Brasil](https://www.bbc.com/portuguese)

[BBC News Mundo (Spanish)](https://www.bbc.com/mundo)

[BBC News မြန်မာ (Burmese)](https://www.bbc.com/burmese)

[BBC News 中文 (Chinese)](https://www.bbc.com/zhongwen/simp)

[BBC News Indonesia](https://www.bbc.com/indonesia)

[BBC News 코리아 (Korean)](https://www.bbc.com/korean)

[BBC News ไทย (Thai)](https://www.bbc.com/thai)

[BBC News Tiếng Việt (Vietnamese)](https://www.bbc.com/vietnamese)

[BBC News Azərbaycanca (Azeri)](https://www.bbc.com/azeri)

[BBC News Magyarul (Hungarian)](https://www.bbc.com/magyarul)

[BBC News Кыргыз Кызматы (Kyrgyz)](https://www.bbc.com/kyrgyz)

[BBC News Polska (Polish)](https://www.bbc.com/polska)

[BBC News România (Romanian)](https://www.bbc.com/romania)

[BBC News Русская служба (Russian)](https://www.bbc.com/russian)

[BBC News na srpskom (Serbian)](https://www.bbc.com/serbian/lat)

[BBC News Україна (Ukrainian)](https://www.bbc.com/ukrainian)

[BBC News O'zbek (Uzbek)](https://www.bbc.com/uzbek)

[BBC News عربي (Arabic)](https://www.bbc.com/arabic)

[BBC News فارسی (Persian)](https://www.bbc.com/persian)

[BBC News Türkçe (Turkish)](https://www.bbc.com/turkce)

[BBC News বাংলা (Bengali)](https://www.bbc.com/bengali)

[BBC News دری (Dari)](https://www.bbc.com/dari)

[BBC News ગુજરાતી (Gujarati)](https://www.bbc.com/gujarati)

[BBC News हिन्दी (Hindi)](https://www.bbc.com/hindi)

[BBC News मराठी (Marathi)](https://www.bbc.com/marathi)

[BBC News नेपाली (Nepali)](https://www.bbc.com/nepali)

[BBC News සිංහල (Sinhala)](https://www.bbc.com/sinhala)

[BBC News پښتو (Pashto)](https://www.bbc.com/pashto)

[BBC News ਪੰਜਾਬੀ (Punjabi)](https://www.bbc.com/punjabi)

[BBC News Afaan Oromoo](https://www.bbc.com/afaanoromoo)

[BBC News አማርኛ (Amharic)](https://www.bbc.com/amharic)

[BBC News Afrique (French)](https://www.bbc.com/afrique)

[BBC News Hausa](https://www.bbc.com/hausa)

[BBC News Gaelic NAIDHEACHDAN](https://www.bbc.com/naidheachdan)

[BBC News Ìgbò](https://www.bbc.com/igbo)

[BBC News Japanese 日本語](https://www.bbc.com/japanese)

[BBC News Gahuza](https://www.bbc.com/gahuza)

[BBC News Pidgin](https://www.bbc.com/pidgin)

[BBC News Somali](https://www.bbc.com/somali)

[BBC News Swahili](https://www.bbc.com/swahili)

[BBC News தமிழ் (Tamil)](https://www.bbc.com/tamil)

[BBC News తెలుగు (Telugu)](https://www.bbc.com/telugu)

[BBC News ትግርኛ (Tigrinya)](https://www.bbc.com/tigrinya)

[BBC News اردو (Urdu)](https://www.bbc.com/urdu)

[BBC News Yorùbá](https://www.bbc.com/yoruba)

[BBC News World Service](https://www.bbc.com/ws/languages)

Follow BBC on:

* [Terms of Use](https://www.bbc.com/pages/terms-of-use) * [Subscription Terms](https://www.bbc.com/pages/subscription-terms) * [About the BBC](https://www.bbc.co.uk/aboutthebbc) * [Privacy Policy](https://www.bbc.com/pages/privacy-policy) * [Cookies](https://www.bbc.com/usingthebbc/cookies/) * [Accessibility Help](https://www.bbc.co.uk/accessibility/) * [Contact the BBC](https://www.bbc.co.uk/contact) * [Advertise with us](https://advertising.bbcstudios.com/) * [Do not share or sell my info](https://www.bbc.com/usingthebbc/cookies/how-can-i-change-my-bbc-cookie-settings/) * [BBC.com Help & FAQs](https://help.bbc.com/hc/) * [Content Index](https://www.bbc.com/pages/content-index) * [Set Preferred Source](https://www.bbc.com/future/article/20260128-how-to-make-google-put-trusted-sources-up-top-when-you-search)

Copyright 2026 BBC. All rights reserved. The BBC is not responsible for the content of external sites.[**Read about our approach to external linking.**](https://www.bbc.com/editorialguidelines/guidance/links-and-feeds)