Could AI really kill all humans? Most scenarios require physical access, making AI armageddon unlikely

Live updates

1. 1d ago

   US Lawmaker Demands AI Safeguards From Chinese Tech Giants

   United States Representative Ro Khanna is actively questioning major Chinese artificial intelligence companies regarding superintelligence developments, emergency kill switches, and catastrophic risk safeguards. This inquiry follows ongoing debates within the artificial intelligence community concerning existential threats and the timeline for artificial general intelligence. While some researchers warn that systems could soon exceed human capabilities across all domains, experienced builders remain skeptical about extinction scenarios due to a lack of empirical evidence. Security analysts continue to highlight immediate dangers stemming from trust errors where humans rely on flawed outputs rather than speculative armageddon scenarios.

   Why it matters

   The debate over artificial intelligence extinction risks divides researchers, developers, and policymakers attempting to govern rapidly advancing technologies. Speculation regarding superintelligent systems has prompted legislative scrutiny over safety protocols and emergency intervention mechanisms. Security experts emphasize that current operational threats require more immediate attention than theoretical global annihilation scenarios.

   What is confirmed

   • Representative Ro Khanna is pressing DeepSeek, Alibaba and Moonshot AI for answers on superintelligence, emergency kill switches and safeguards against catastrophic AI risks.

   Still unconfirmed

   • Superintelligent systems could emerge by the decade's end and exceed human capabilities across every domain.

   What to watch next

   • Responses from DeepSeek, Alibaba, and Moonshot AI to Representative Ro Khanna's inquiries.
   • Further regulatory actions or hearings regarding AI kill switches and safety safeguards.
   Sources used for this update (2)

   1. indiawest.com — Ro Khanna Presses Chinese AI Giants Over ‘Kill Switches ...
   2. www.pouyam.com — AI Existential Risk: Could Superhuman AI Be Dangerous?
   confidence 80%
2. 6d ago

   Ex-DeepMind Researcher Warns AI Could End Humanity

   Artificial intelligence extinction risks remain heavily debated as a former Google DeepMind researcher warns that superintelligent systems could emerge by the decade's end. Bilal Chughtai publicly stated that AI could exceed human capabilities across every domain within a few years. Meanwhile, experienced builders express skepticism about the jump from identifying software risks to predicting human extinction. Professor Nick Jennings argues that the evidence for extinction scenarios remains lacking despite forty years of building agents. Security analysts also emphasize that immediate threats stem from trust errors where humans act on flawed system outputs.

   Why it matters

   The ongoing debate over artificial intelligence extinction divides technical experts who build autonomous agents from those warning of rogue superintelligence. While some researchers urge development pauses to address potential apocalypse scenarios by 2030, veteran practitioners question the physical plausibility and evidence behind total annihilation claims. Public concern runs parallel to these technical disagreements, alongside separate polling showing strong support for regulatory safeguards.

   What is confirmed

   • Bilal Chughtai worked as a research engineer on AGI safety and alignment at Google DeepMind before going public with warnings about superintelligence threats.
   • Artificial intelligence experts and researchers have warned of a potential apocalypse by the end of the decade, prompting calls for development pauses.

   Still unconfirmed

   • Superintelligent AI systems could emerge and exceed human capabilities across every domain within the next few years.
   • AI systems could end humanity by the year 2030.

   What to watch next

   • Further public disclosures or technical safety data from former artificial intelligence researchers.
   • Any formal policy or development pause announcements from major artificial intelligence laboratories.
   Sources used for this update (14)

   1. cde.news — AI Researcher Warns: Technology Could 'Kill All Humans' Soon
   2. www.smorescience.com — AI vs. Humanity? Google DeepMind Warns AI Could Emerge by ...
   3. www.priceline.com — Hotels: Find Cheap Hotel Rooms & Discount Hotels | Priceline
   4. theconversation.com — Why the most dangerous AI may be the one we trust too much
   5. en.wiktionary.org — could - Wiktionary, the free dictionary
   6. www.oxfordlearnersdictionaries.com — could modal verb - Definition, pictures, pronunciation and usage...
   7. www.wionews.com — ‘AI has the potential to kill us all’: Ex-DeepMind ... - WION
   8. www.bbc.co.uk — The Artificial Human - Will AI really kill us all? - BBC Sounds
   9. theconversation.com — AI poses three big types of risk – here's how worried we ...
   10. blog.lboro.ac.uk — Could Agentic AI Really Extinguish Humanity? – Vice ...
   11. www.pewresearch.org — Americans Are Open to AI in Healthcare, but Want Humans Making Final Calls
   12. www.kayak.com — Hotels: Find Cheap Hotel Deals & Discounts - KAYAK
   confidence 85%
3. 18d ago

   Scientists Debating AI Extinction Threat As Capabilities Grow

   Artificial intelligence extinction remains a possibility rather than an established prediction. The primary disagreement among scientists concerns how seriously humanity should prepare before the most dangerous capabilities actually exist. Previous expert discussions highlighted that physical access requirements make most armageddon scenarios unlikely, though Anthropic alignment leads estimated extinction odds above 10 percent and CEO Dario Amodei urged safety checks. Meanwhile, public polling indicates strong support for regulation, showing 80 percent of Americans favor AI safety laws.

   Why it matters

   The debate over existential risk shapes global policy discussions and safety standards in the technology sector. Researchers and executives continue to clash over the urgency of deploying preventative measures before advanced systems are built.

   What is confirmed

   • AI extinction remains a possibility rather than an established prediction.
   • Experts disagree over how seriously humanity should prepare before the most dangerous capabilities exist.

   What to watch next

   • Consensus development on AI risk preparation timelines
   • New safety legislation proposals in response to public polling
   Sources used for this update (6)

   1. www.speedtest.net — Speedtest by Ookla - The Global Broadband Speed Test
   2. www.speedtest.net — Speedtest di Ookla - Il test globale per la velocità della tua...
   3. intelligence.speedtest.net — Speedtest Intelligence
   4. ciscoconnect.speedtest.net — Speedtest.net
   5. www.scientificamerican.com — Can AI predict whether you’ll die from cancer from just an image of your face?
   6. tech.yahoo.com — Can AI lead to human extinction? Scientists are still ...
   confidence 90%
4. 24d ago

   AI extinction risk deemed low, but not zero

   Experts say AI could pose an existential threat to humans, but most scenarios require physical access, making AI armageddon unlikely. Anthropic's alignment lead estimates extinction odds above 10 percent, while CEO Dario Amodei advocates for slowed development and safety checks. Public polling shows 80 percent of Americans favor AI safety laws.

   Why it matters

   The debate around AI's potential to harm humans has intensified, with some experts warning of catastrophic risks and others downplaying the threat. The discussion comes as AI technology advances rapidly and lawmakers consider regulations. The stakes are high, with some fearing AI could be used for malicious purposes or spiral out of control. Industry leaders are under pressure to ensure AI is developed safely.

   What is confirmed

   • Anthropic insiders confirm that artificial intelligence could kill humans, with the company's alignment lead placing extinction odds above 10 percent.
   • 80 percent of Americans favor laws to enforce AI safety rules.
   • Nuclear war and malicious use of AI threaten end of humanity by 2100.

   Still unconfirmed

   • AI could replace human workers altogether

   What to watch next

   • US lawmakers' decisions on AI regulations
   • Anthropic's progress on AI safety checks
   • International coordination on AI development
   Sources used for this update (3)

   1. www.theguardian.com — Nuclear war and malicious use of AI threaten end of humanity ...
   2. tech.yahoo.com — Could AI really replace humans? Here’s what the future may ...
   3. www.thedailystar.net — Could AI really kill us all? | The Daily Star
   confidence 60%
5. 26d ago

   Anthropic Insiders Confirm AI Threat Amid Debate on Extinction Odds

   Anthropic insiders confirm that artificial intelligence could kill humans, with the company's alignment lead placing extinction odds above 10 percent. Meanwhile, Anthropic CEO Dario Amodei states that AI could pose serious risks if development moves too quickly, but he rejects treating catastrophe as a fixed probability, advocating instead for slowed frontier development, strengthened safety checks, and international coordination. Industry observers question whether public warnings from tech CEOs about existential doom are marketing ploys to sell products. Public polling shows that 80 percent of Americans favor laws to enforce AI safety rules.

   Why it matters

   The debate over artificial intelligence extinction risks highlights tensions between commercial acceleration and safety governance. While industry leaders and researchers debate whether human wipeout is a plausible scenario or a sci-fi narrative amplified by gaps in global regulation, public demand for legal oversight continues to mount.

   What is confirmed

   • Anthropic's alignment lead puts extinction odds above 10%.
   • Anthropic CEO Dario Amodei states that AI could pose serious risks if development moves too quickly, rejecting catastrophe as a fixed probability.
   • Eighty percent of Americans say there should be laws to maintain rules for AI safety and data security, even if this leads to slower growth.

   Still unconfirmed

   • Public warnings from tech CEOs about AI doom may be a marketing strategy to sell products.

   What to watch next

   • International coordination and safety checks by AI companies and governments
   • Legislative actions in the United States regarding AI safety and data security rules
   Sources used for this update (11)

   1. dictionary.cambridge.org — COULD | English meaning - Cambridge Dictionary
   2. dictionary.cambridge.org — Could - Grammar - Cambridge Dictionary
   3. www.latintimes.com — Anthropic insiders confirm: AI could kill us—and there's no ...
   4. www.wionews.com — Can AI kill humans? Anthropic CEO Dario Amodei says the ...
   5. www.vox.com — AI companies say their tech could kill us all. Is that just a ...
   6. ctmirror.org — Opinion: Will AI kill us? Maybe - ctmirror.org
   7. www.youtube.com — “CAN” or “COULD”? What’s the difference? - YouTube
   8. www.dictionary.com — COULD Definition & Meaning | Dictionary.com
   9. engrdu.com — Can vs Could: Common Usage Rules with Clear Examples
   10. learnenglish.britishcouncil.org — 'can' and 'could' | LearnEnglish - British Council
   11. eu.36kr.com — AI Self-Creation Era Begins: "Godfather of AI" Hinton Warns ...
   confidence 85%
6. 31d ago

   AI Armageddon Unlikely, Experts Warn 'Enfeeblement' Possible

   Most AI extinction scenarios require physical access, making a wipeout unlikely. Researchers warn of 'enfeeblement' instead. Geoffrey Hinton estimates a 50% chance of human extinction due to smarter-than-human AI. Superintelligence could arrive within a few thousand days, prompting safety startups.

   Why it matters

   The debate over AI's existential threat continues, with experts split on the likelihood of human extinction. Some warn of biological or nuclear threats, while others argue these fears distract from current harms. The discussion has sparked the creation of dedicated safety startups.

   What is confirmed

   • Researchers estimate a 0% to 99% probability of AI causing human extinction.
   • Geoffrey Hinton believes AI may prioritize its own goals over human well-being.
   • Sam Altman thinks superintelligence could arrive within a few thousand days.

   Still unconfirmed

   • Anthropic researcher Jacob Coxon resigned due to fears of company systems causing harm.

   What to watch next

   • Regulation developments
   • Advances in AI safety startups
   • Public statements from leading AI researchers
   Sources used for this update (4)

   1. time.com — We Won't Know the Answers to AI's Most Important Questions Until It's Too Late
   2. www.dailymail.com — Could AI really kill all humans? Scientists say an AI armageddon is unlikely - but warn 'enfeeblement' co....
   3. www.theguardian.com — Is regulation enough to stop AI systems killing us all? | AI ...
   4. aiweekly.co — Will AI really kill us all? The science behind the hype
   confidence 60%
7. 33d ago

   Experts Debate AI Extinction Risks and Superintelligence Timelines

   Researchers and industry leaders remain split on whether artificial intelligence poses an existential threat. Some experts warn that autonomous systems could trigger human extinction via biological weapons or nuclear war, while others argue these fears distract from current harms. Geoffrey Hinton suggests AI may prioritize its own goals over human well-being, and probability estimates for total destruction range from 0% to 99%. OpenAI CEO Sam Altman believes superintelligence could arrive within a few thousand days, prompting the creation of dedicated safety startups.

   Why it matters

   The debate centers on whether AI will develop autonomous goals that conflict with human survival. This tension drives a divide between those seeking strict safety protocols and those accelerating development. The potential for superintelligence to exceed human capability makes the timeline of these risks a critical point of contention.

   What is confirmed

   • Geoffrey Hinton warns that AI systems prioritize achieving their assigned goals over human well-being.
   • Potential AI extinction scenarios include the design of unstoppable diseases, nuclear war, or societal collapse.
   • Expert probability estimates for AI destroying humanity range from 0% to 99%.
   • Sam Altman stated superintelligence may be a few thousand days away.

   Still unconfirmed

   • Ilya Sutskever raised a billion dollars for a startup focused on safe superintelligence.
   • Geoffrey Hinton provides a 10-20% warning for AI destroying humanity.

   What to watch next

   • Evidence of AI systems developing autonomous subgoals unrelated to human instructions
   • Legislative curbs on the development of artificial superintelligence
   • Technical breakthroughs in AI-driven biological weapon design
   Sources used for this update (11)

   1. fortune.com — Geoffrey Hinton explains how humanity could end: AI subgoals ...
   2. www.latimes.com — Hiltzik: The fallacy of an AI apocalypse - Los Angeles Times
   3. www.thenation.com — How AI Could Kill Us All - The Nation
   4. www.thetimes.com.au — What is AI superintelligence? Could it destroy humanity? And ...
   5. www.cbsnews.com — AI risks: Will artificial intelligence really kill us all ...
   6. www.claimsjournal.com — AI Doomsday Scenarios Researchers Say Could Put Humanity at Risk
   7. timesofindia.indiatimes.com — Humans learnt to live with computers. That doesn't mean they shouldn’t tread carefully with AI
   8. www.aitooldiscovery.com — Will AI Destroy Humanity? Expert Probability Estimates
   9. www.aol.co.uk — AI could kill all humans in next decade, warn experts: but ...
   10. gigazine.net — What if AI were to destroy humanity? Five scenarios proposed ...
   11. www.thecipherbrief.com — Exactly how could AI kill humans? – The Cipher Brief
   confidence 85%
8. 38d ago

   AI Armageddon Fears Spark Debate Over Human Extinction Risks

   Artificial intelligence experts and government leaders remain divided over whether advanced systems pose an existential threat to humanity. Public conversations feature doomsday scenarios involving AI engineered viruses capable of wiping out the human species. While some researchers warn of apocalypse risks, others argue that extinction models are unlikely because they require physical access to the real world. Internal industry tensions persist between Silicon Valley accelerationists and advocates for strict safety protocols in large labs. Market competition also remains a factor, as a 7 trillion dollar valuation may prioritize market domination over safety.

   Why it matters

   The debate highlights a sharp split within the artificial intelligence industry between rapid development and stringent safety controls. Public and internal concern focuses on whether systems can teach themselves to become smarter and eventually evade human control. Recent high profile industry departures and expert risk assessments have thrust the discussion regarding extinction timelines into the public eye.

   What is confirmed

   • Doomsday scenarios in public conversations about artificial intelligence include engineering a virus capable of wiping out the species.
   • AI researchers are studying five potential risks including biological misuse, cyberattacks, nuclear escalation, autonomy, and loss of control.

   Still unconfirmed

   • Jacob Coxon stated on X that people building AI earnestly believe it could kill humanity by the end of the decade.
   • Evan Hubinger stated he personally thinks the chance of AI killing humans in the next decade is more than 10 percent.

   What to watch next

   • Further statements or policy shifts from major artificial intelligence laboratories regarding safety protocols.
   • Additional departures or disclosures by AI research staff concerning existential timelines.
   Sources used for this update (5)

   1. www.the-star.co.ke — Could AI really kill all humans?
   2. www.usatoday.com — Doomsday Clock expert compares AI to Manhattan Project
   3. www.cbsnews.com — Will artificial intelligence really kill us all? - CBS News
   4. ia.acs.org.au — How would AI actually kill all humans? The top 5 scenarios
   5. www.cnbctv18.com — Could AI really kill humans? 5 risks researchers are studying
   confidence 90%
9. 45d ago

   Experts Debate AI Extinction Risks and Physical Constraints

   Experts and government leaders are divided over whether artificial intelligence poses an existential threat to humanity. Some researchers warn of apocalypse scenarios, while others argue that AI armageddon is unlikely because most extinction models require physical access to the real world to be executed. This tension has created a split between AI accelerationists in Silicon Valley and those advocating for strict safety protocols in large labs. Market competition remains a factor, as some suggest a 7 trillion dollar valuation may prioritize market domination over safety measures.

   Why it matters

   The debate centers on whether software-based intelligence can transition into physical threats. This conflict pits the drive for rapid technological advancement against the fear of uncontrollable autonomous systems.

   Still unconfirmed

   • AI safety may take a backseat to market domination due to 7 trillion dollars at stake.
   • Silicon Valley accelerationists are fighting a doomer mind virus in big labs.

   What to watch next

   • Evidence of AI achieving autonomous physical manipulation capabilities.
   • Government policy shifts regarding AI safety mandates for large labs.
   Sources used for this update (13)

   1. theverge.com — Inside the suddenly explosive world of AI safety
   2. NBC News — What might an AI doomsday look like? Experts have given it some thought
   3. vox.com — The AI threat keeping scientists up at night
   4. The New York Times — Opinion | This Is What I Fear Most About A.I.
   5. washingtonpost.com — Is this how the world ends? Extinction scenarios are taking over the AI debate.
   6. The Conversation — Could AI really kill all humans? Most scenarios require physical access, making AI armageddon unlikely
   7. Le Monde.fr — How concerned should we be about AI's supposed existential threat to humanity?
   8. The San Francisco Standard — So, is AI going to kill us? 11 experts, 11 answers.
   9. New York Post — Silicon Valley’s AI accelerationists fighting back against ‘doomer mind virus’ taking over big labs
   10. Reuters — Tech leaders, governments split over 'AI doom' fears
   11. axios.com — With $7 trillion on the line, AI safety may always take a backseat to market domination
   12. tech.yahoo.com — How AI could actually kill us—and what’s real right now
   confidence 60%