I thought this might make an interesting topic for discussion.
Over the last few days there has been quite a lot of coverage in respected mainstream news sources about concerns being raised by researchers actually working at some of the leading AI companies.
What particularly caught my attention was a comment from Evan Hubinger, an alignment researcher at Anthropic, who said:
“We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade.”
Obviously that doesn't mean scientists have calculated that there is literally a 10% probability that AI will wipe out mankind. This is Hubinger's personal assessment of the risk, and probabilities involving something unprecedented like superintelligent AI are clearly extremely difficult to quantify.
But I do find it remarkable that people working near the forefront of this technology are openly talking in these terms — and that Reuters, the Financial Times, Washington Post, Wall Street Journal, Guardian and CNN are now treating the subject seriously.
Some of the recent coverage:
Reuters — “More US lawmakers seek new AI rules after Anthropic researchers warn of human extinction”
https://www.reuters.com/business/openai-faces-senate-probe-into-hugging-face-incident-axios-reports-2026-09-10/
Reuters reports that the warnings from Anthropic researchers have now prompted calls from US lawmakers for stronger AI safety regulation.
The Guardian — “Anthropic researchers say AI could cause human extinction by 2030”
https://www.theguardian.com/technology/2026/sep/09/anthropic-researchers-ai-human-extinction
This covers the comments from Jacob Coxon, Evan Hubinger and Samuel Marks. Hubinger gives his personal estimate as greater than 10% within the next decade.
Washington Post — “Political world erupts as AI researchers warn of ‘extinction’ threat”
https://www.washingtonpost.com/technology/2026/09/09/anthropic-researcher-resigns-warning-reckless-race-toward-superintelligence/
The Washington Post reports Hubinger's >10% estimate and discusses the fact that Anthropic has deliberately built a reputation for being one of the more safety-conscious AI companies.
Financial Times — “Anthropic researcher quits over AI labs ‘gambling with our lives’”
https://www.ft.com/content/20c07191-8da6-440f-b04b-8ea0ebdd9153
The FT reports on researcher Jacob Coxon's resignation from Anthropic and his concern that companies are racing towards increasingly powerful, potentially self-improving AI without having solved the problem of reliably controlling it.
Wall Street Journal — “How Would AI Actually Kill Us All? What to Know About the AI Doomsday Debate”
https://www.wsj.com/tech/ai/how-would-ai-actually-kill-us-all-what-to-know-about-the-ai-doomsday-debate-034270d1
The WSJ looks at what researchers actually mean when they talk about an existential AI risk — things such as losing control of very capable autonomous systems or AI dramatically increasing people's ability to create biological or other catastrophic weapons.
CNN — discussion of the Anthropic warnings
https://transcripts.cnn.com/show/cg/date/2026-09-09/segment/01
CNN also reports Hubinger's greater-than-10% estimate, while importantly pointing out that any precise percentage should be regarded as a judgment rather than a measured probability.
Of course, from a Scriptural point of view, we don't believe that AI — or anything else humans develop — is ultimately going to frustrate Jehovah's purpose for mankind and the earth. Jehovah says that he formed the earth “to be inhabited” (Isaiah 45:18), and Psalm 37:29 says that “the righteous will possess the earth, and they will live forever on it.”
So I'm certainly not suggesting that AI is actually going to exterminate humanity.
What interests me is the contrast: mankind is developing technology so powerful that some of the people closest to it are now publicly discussing the possibility of losing control of it, to the point that human extinction is appearing in mainstream political and scientific discussion.
Only a few years ago that would probably have sounded like science fiction.
What do you make of these warnings?
Do you think researchers at the forefront of AI are seeing a genuine danger that most people haven't appreciated yet?
Or is putting figures such as “10%” on something this uncertain likely to exaggerate a risk that nobody can really calculate?