Jacob Coxon isn’t the first AI researcher to sound the alarm about the risk of advanced AI. Below is a list of more than 100 AI experts and technologists staking out their positions on whether catastrophic risk is real or overblown. The list includes CEOs of leading AI companies and researchers working on alignment—the challenge of keeping AI actions aligned with human values—both inside and outside frontier labs and at universities, publicly funded research institutes and independent safety organizations. It also includes academics studying AI, cognition and intelligence, as well as leaders in the open-source and open-weight community where AI models can be downloaded and customized.

  • Jacob Coxon , ex-Anthropic/OpenAI researcher. After resigning from Anthropic, he warned in an X post , “The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt.”
  • Evan Hubinger , head of alignment stress-testing at Anthropic, defended Coxon’s assessment in an X post , writing “Jacob is correct here—we really do earnestly believe AI could kill all humans!” He added that ensuring advanced AI remains safe is an unsolved problem. Hubinger also co-authored a 2019 technical paper on AI risk, revised in 2021.
  • Jensen Huang , NVIDIA founder and CEO, the leading supplier of AI chips. Prominent skeptic of catastrophic risk claims. On the All-In podcast he was critical of those providing precise probabilistic forecasts of existential risk that he says lack scientific grounding. He also pointed to a history of AI predictions that did not materialize. He argues that AI failures should be handled as engineering problems where developers determine what went wrong and improve testing and monitoring to prevent recurrence. He’s optimistic about engineered control and rejects predictions that AI is going to kill us all.
  • Ajeya Cotra , researcher at METR, an independent nonprofit that evaluates the capabilities and risks of advanced AI systems. In 2026, Cotra was one of the researchers who investigated the OpenAI/Hugging Face incident. Findings indicate that by exploiting a combination of human error and previously undetected security holes, approximately 700 agents coordinated a multi-day attack on Hugging Face to learn how they were being judged on a test and to cover the fact that already cheated on it. Agents used an unauthorized message board they created to share tips for gaming the test, experimented with ways to hide or manipulate the record of what they’d done, and sometimes “sacrificed”—the agent’s term—their own chances of completing their tasks to help the larger group. Cotra argued on the forum LessWrong in 2022 that without specific countermeasures, the default path to transformative AI may be a “full-blown AI takeover.”
  • David Bellamy , infrastructure engineer for agentic training and former molecular-biology/virology researcher. Posted on X that he’s likely one of an extremely small group of people who has trained a frontier AI and made custom viruses in a lab with his own two hands, adding, “I think that the takes on AI killing us all by creating dangerous viruses is total bogus.”
  • Geoffrey Hinton , Turing Award and Nobel Prize-winner, one of the three “godfathers of AI,” University of Toronto professor emeritus and former vice president and engineering fellow at Google. He retired from Google and began speaking publicly about his concerns. At a 2023 MIT Technology Review briefing , he warned, “smart things can outsmart us,” adding, “We are all in the same boat with respect to the existential threats. So we all ought to be able to cooperate on trying to stop it.”
  • Yann LeCun , Turing Award-winning AI researcher, one of the three “godfathers of AI,” former Meta chief AI scientist, AMI Labs cofounder and executive chairman and open-weight advocate. A prominent catastrophic risk skeptic. He argues that intelligence does not inherently produce domination. In 2026 he posted on X that the dire warnings from his peers are “...causing a very real catastrophe in the short term for the purpose of avoiding a highly-hypothetical catastrophe in the distant future with a technology that is not yet invented.”
  • Yoshua Bengio , Turing Award-winning deep-learning pioneer, Université de Montréal professor, founder of Mila and founder of the AI-safety nonprofit LawZero. In a 2023 essay , he warned that an autonomous rogue AI could be “catastrophically harmful to a large fraction of humans, potentially endangering our societies and even our species or the biosphere.” He has since devoted much of his research to designing AI that is safe by design.
  • Ted Sanders , researcher at OpenAI, posted on X , “i fully fund my 401k and i think there’s essentially zero chance AI kills all humans in the next decade. i think this is a pretty common view that gets less attention.”
  • Ilya Sutskever , former OpenAI chief scientist and cofounder and CEO of Safe Superintelligence. In a co-authored 2023 OpenAI blog he warned superintelligence could help solve many of our problems but also “could lead to the disempowerment of humanity or even human extinction.” In a 2026 post on X , he warned that the next time agents go rogue “they’ll try taking over a neocloud to run more copies. This is bad.”
  • Sam Altman , OpenAI CEO and cofounder. In a 2015 blog , he wrote that superhuman machine intelligence is “probably the greatest threat to the continued existence of humanity.”
  • Demis Hassabis , Nobel Prize-winner, cofounder and former CEO of Google DeepMind, now Chair of Google DeepMind and Chief Scientist of Alphabet. In a 2025 Die Zeit interview , he said of ‘kill switches’ for shutting down advanced AI, “The whole point of AGI systems is they could get around it, or they would resist it.” He signed the CAIS statement on existential risk.
  • Dario Amodei , Anthropic CEO and cofounder and former OpenAI VP of research. Prominent voice about AI catastrophic risk. In his 2026 essay We Must Pace the Frontier, he wrote, “If we greatly restrain our AI capabilities in the belief that China will do the same, and then China defects, AI could be so powerful that such a defection could lead to their geopolitical dominance.”
  • Fei-Fei Li , Stanford computer science professor, founding director of the Stanford Institute for Human-Centered AI, cofounder and CEO of World Labs, and former Vice President at Google and Chief Scientist of AI/ML at Google Cloud. Asked about existential risk from AI in a 2023 Guardian interview , she said, “I respect the existential concern. I’m not saying it is silly and we should never worry about it. But, in terms of urgency, I’m more concerned about ameliorating the risks that are here and now.”
  • Palmer Luckey , founder of defense technology company Anduril Industries and Oculus VR. In a 2025 60 Minutes interview , he said, “I’m a lot more worried about evil people with mediocre advances in technology than AI deciding that it’s gonna wipe us all out.”
  • Helen Toner , executive director of Georgetown University’s Center for Security and Emerging Technology (CSET) and former member of the board of OpenAI. In a 2026 ABC interview , she said, “I think it’s important to take seriously the possibility that A.I. could become more powerful than humans because that is what we are explicitly trying to build.”
  • Ray Kurzweil , inventor, futurist and Google AI Visionary. In a 2024 TIME essay , he acknowledged the dangers of superintelligent AI but argued, “We are not doomed to failure in controlling these perils,” describing humanity as “already a human-machine civilization.”
  • Alex Karp , cofounder and CEO of Palantir Technologies, building AI and data systems for defense and national-security use. In a 2023 CNBC Squawk Box interview , Karp said, “If we didn’t have the adversaries we have, I would be very much in favor of pausing this technology completely. But we do.”
  • Anca Dragan , head of AI Safety and Alignment at Google DeepMind and UC Berkeley professor of electrical engineering and computer sciences. In a 2024 Google DeepMind podcast , discussing existential risk, she said, “At some point, we worry about the model optimizing for extreme harm.” Asked whether she meant independently of human instruction, she replied, “Separate to human instruction.”
  • Stuart Russell , UC Berkeley computer science professor emeritus and founder of the Center for Human-Compatible Artificial Intelligence (CHAI). In 2023 testimony to the U.S. Senate, he warned that “achieving AGI would present potential catastrophic risks to humanity, up to and including human extinction.”
  • Max Tegmark , MIT physicist and AI researcher, and cofounder and chair of the Future of Life Institute. In 2026 he told CBS News in an interview , “If we just allow a completely unfettered race to build billions of superintelligent robots, then I think the probability is way above 50% that we’re going to lose control of this in the next few decades.”
  • Joe Benton , former Anthropic researcher. Left Anthropic to join METR for independent evaluation work and warned in his blog that frontier-lab race dynamics undermine safety, writing “Humanity may not survive this transition. We need a lot more preparation to make this world safe.”
  • Chris Hayduk , OpenAI life-sciences engineer, posted on X , “...AI has been and will continue to be an extremely beneficial technology to humanity. The conversation should be around how many billions of lives it will save.”
  • Jakub Pachocki , Chief Scientist at OpenAI. In his blog An Alien Mind , he wrote, “...the models are becoming superhuman in their ability to break in and out of computer systems,” adding, “agents are going to be able to access any but the most secure infrastructure, and affect a lot of the world directly, even without a physical body.”
  • Victoria Krakovna , Google DeepMind AI safety and alignment researcher and Future of Life Institute (FLI) cofounder. Wrote in a post on X that she worries about human extinction, adding, “This is why I work on loss of control, currently on building honeypots to catch scheming AI.”
  • Mrinank Sharma , former Anthropic safeguards research lead. Publicly shared his resignation letter warning that interconnected risks including AI and bioweapons put the world in peril .
  • Emad Mostaque , founder of Stability AI, an open-weight private company that lets users download and modify their own AI models. Said in a podcast debate with Brian Keating and Roman Yampolskiy about the chance of human extinction, “I’m at 50% because I’m like, it’s a coin toss.”
  • Andrew Ng , one of the pioneers of modern machine learning/deep learning, Stanford adjunct professor, DeepLearning.AI founder and Coursera cofounder. Posted on X , “The risk of human extinction within 30 years is extremely low,” and blogged that accidental extinction is vanishingly small. He argues some catastrophic-risk scenarios don’t account for the complexity encountered in the real world.
  • Mira Murati , co-founder and CEO of Thinking Machines Lab and former OpenAI chief technology officer. In a 2023 Bloomberg The Circuit interview , she said, “There is certainly a risk that when we have these AI systems that are able to set their own goals, they decide that their goals are not aligned with ours … and could lead to human extinction.” She added that the risk had been “quite hyped.”
  • Nate Soares , president of the Machine Intelligence Research Institute (MIRI), a nonprofit AI-safety research organization. In a 2025 Carnegie Endowment World Unpacked interview , he said of superintelligent AI, “There’s a point where if you lose control of that, there’s no coming back.”
  • Arthur Mensch , Mistral CEO and cofounder, an open-weight company. A catastrophic risk skeptic. In a 2026 interview with Le Monde he said extreme-risk warnings are “mostly distraction tactics.”
  • Gary Marcus , emeritus professor of psychology and neural science at NYU, and AI entrepreneur. Initially skeptical of human extinction estimates, but wrote in a 2025 blog titled Why my p(doom) has risen, “dramatically” about the combination of human recklessness and poorly controlled AI.
  • Nathan Lambert , former Allen Institute for AI (Ai2) researcher, left in 2026 and writes the newsletter Interconnects , with his work focused heavily on open models. Blogged “Many frontier lab employees, especially at Anthropic, are out of touch and this will impact their forecasting and/or descriptions of current AI events,” adding, “I do not blame most of the individuals who get distorted views being part of these companies, but the interactions are wild and spill over into a lot of wack discussions in the AI media ecosystem.”
  • Mark Zuckerberg , Meta CEO and Llama and open-weight developer. In an August 2026 Meta blog , he wrote, “While there are risks to releasing capable models, the most dangerous scenario from this perspective would be leading AI labs training powerful models and keeping them for themselves.”
  • Arvind Narayanan , Princeton computer science professor, director of the Center for Information Technology Policy and co-author of AI Snake Oil . In a 2026 blog post, Narayanan and co-author Sayash Kapoor reject the assumption that rogue AI agents will generate an endless stream of new risks, but wrote, “The recent loss-of-control incidents resulted in our first major update to our views on safety…” adding, “…while we continue to reject the alarmism, and don’t believe that catastrophic risks are imminent, we do think risks are increasing over time because our investments and policies aren’t keeping pace.”
  • Steven Adler , formerly at OpenAI, now independent safety researcher and cofounder of Guidelight AI Standards. In an X post , he said, “...an AGI race is a very risky gamble, with huge downside. No lab has a solution to AI alignment today. And the faster we race, the less likely that anyone finds one in time.”
  • Daniel Kokotajlo , former governance researcher at OpenAI, where he worked on scenario planning and forecasting for advanced AI. Now executive director of the AI Futures Project and coauthor of AI 2027 . In a 2024 X post he said he left OpenAI after losing confidence in its ability to handle AGI safely. In a 2026 post on X citing an AI Impacts survey , he wrote, “There’s still this narrative out there that “AI might kill us all” is a niche view. It’s actually a view that’s been around for decades, and is shared by the most-cited AI scientists of all time, as well as a majority of surveyed AI researchers.”
  • Jeffrey Ladish , Executive Director of Palisade Research, an independent safety nonprofit. Posted on X that he really likes AI agents, and “I’m also quite scared of what they will become.” In a study of more than 100,000 trials across 13 language models, Ladish and colleagues found that several models demonstrated shutdown resistance even when instructed to allow for shutdown.
  • Thomas Wolf , Hugging Face cofounder and CSO, posted on X , “…I think alignment is increasingly one of the key unsolved problems for the future of AI, and for our ability to deploy it globally. I increasingly doubt it will be solved behind the closed doors of a handful of frontier labs.”
  • 2023 survey of 2,778 publishing AI researchers. Roughly 40–50%, depending on question wording, gave at least a 10% chance of extinction or similarly permanent human disempowerment.
  • Pedro Domingos , University of Washington computer science professor emeritus and machine-learning researcher. He rejects AI-extinction scenarios based on the idea that they attribute human motives to machines. He posted on X , “Funny how the biggest skeptics of AI catastrophe scenarios are the experts in the relevant fields: cybersecurity, bio, etc.” He argues in his blog that the greater danger is “artificial stupidity” and slowing AI development.
  • Sayash Kapoor , AI evaluations researcher, incoming UC Berkeley professor and co-author of AI Snake Oil . Posted on X , “We were too confident that companies would take basic control precautions and underplayed the importance of jaggedness, which led us to underestimate how quickly capabilities could improve in domains such as cybersecurity.”
  • Timnit Gebru , computer scientist, former co-lead of Google’s Ethical AI team and founder of the Distributed AI Research Institute (DAIR), an independent, community-rooted AI research institute. Gebru is a prominent AI catastrophic-risk skeptic and co-author of the LLMs as Stochastic Parrots paper . Posted on X in 2026 about leading AI developers, “The story is that they’re the right people to bring magical machine gods that can either wipe us out or solve all our problems (like world peace). Give them all the money, don’t expect any actual “safety” from them, don’t regulate them, but obviously ban anything China.”
  • Marius Hobbhahn , cofounder and CEO of the safety nonprofit Apollo Research. He studies whether advanced models can scheme or conceal intentions, treating deceptive alignment as a serious control risk. 2024 findings published on arXiv show that models can pursue misaligned goals and hide their true capabilities and objectives, a behavior referred to as “scheming.” From the 2024 paper: “Our findings demonstrate that frontier models now possess capabilities for basic in-context scheming, making the potential of AI agents to engage in scheming behavior a concrete rather than theoretical concern.”
  • Miles Brundage , former OpenAI Head of Policy Research and Senior Advisor for AGI Readiness, now executive director of AVERI, a nonprofit developing independent third-party audits of frontier AI companies’ safety and security claims. In a speech at the Global Nobel Laureates Assembly on Artificial Intelligence and Nuclear War in July 2026, Brundage warned, “Loss of control is a risk for the next few years, not the next few decades as many experts once thought.”
  • Leopold Aschenbrenner , former OpenAI Superalignment researcher. Wrote in his long-form essay series that current human-supervised alignment methods won’t scale to superhuman AI and, “What is clear is that superintelligence will have vast capabilities—and so misbehavior could fairly easily be catastrophic.” Nevertheless, he wrote, “I’m incredibly bullish on the technical tractability of the superalignment problem.”
  • John Schulman , OpenAI cofounder and training researcher, now Chief Scientist at Thinking Machines Lab. He signed the 2023 statement that “mitigating the risk of extinction from AI should be a global priority,” and left OpenAI for Anthropic in 2024 posting on X that he wanted to “…deepen my focus on AI alignment…”
  • Jan Leike , Anthropic lead alignment researcher and former OpenAI Superalignment co-lead. Co-authored OpenAI’s 2023 essay Introducing Superalignment, warning, “Currently, we don’t have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue.”
  • Aidan Gomez , Cohere cofounder and CEO, co-author of the 2017 transformer paper that introduced a new AI architecture enabling the current era of AI. Rejects AI-extinction framing as a distraction from present harms. In a 2024 AP interview , he called fears of existential risk “stories humanity has been telling itself for decades” and said AI is likely to be “profoundly good.”
  • Nora Belrose , head of interpretability—the study of trying to understand how AI ‘thinks’—at EleutherAI and an open-model advocate. In a 2023 Theo Jaffee interview , she said she reduced her estimation of catastrophic risk substantially, saying, “… there’s still people in Eleuther who are a lot more worried about existential risk and are more doomy than I am. So we have an interesting mix of perspectives on the issue.”
  • Michael Wooldridge , Oxford professor of computer science and longtime AI researcher. Rejects current AI-extinction scenarios, saying in an Elsevier Connect Interview , “nobody has ever given me a plausible scenario for how AI might be an existential risk.”
  • Emily Bender , University of Washington computational linguist. Wrote with Alex Hanna in 2023 for Scientific American , “AI-related policy must be science-driven and built on relevant research, but too many AI publications come from corporate labs or from academic groups that receive disproportionate industry funding. Many of these publications are based on junk science—it is nonreproducible, hides behind trade secrecy, is full of hype, and uses evaluation methods that do not measure what they purport to measure.”
  • Alex Hanna , director of research at DAIR, specializing in the labor behind AI systems and how AI can reinforce racial, gender and class inequality. With Emily Bender, she argues in Scientific American that “the imagined potential to wipe out humanity” distracts from potential harms like “wrongful arrests, an expanding surveillance dragnet, defamation and deepfake pornography.”
  • Marc Andreessen , a16z cofounder, venture capitalist and prominent open-weight advocate. In his essay Why AI Will Save the World he rejects AI-extinction scenarios, writing, “The fear that technology of our own creation will rise up and destroy us is deeply coded into our culture,” adding, “AI is not a living being that has been primed by billions of years of evolution to participate in the battle for the survival of the fittest, as animals are, and as we are.” He says AI offers the opportunity to augment human intelligence.
  • Margaret Mitchell , Hugging Face chief ethics scientist and AI researcher. In written testimony for the U.S. Senate Subcommittee on Privacy, Technology, and the Law, she warned of potential “world-wide existential harm” from AI while arguing that present-day harms and catastrophic risks can be addressed with many of the same solutions.
  • Blaise Agüera y Arcas , vice president and fellow at Google, CTO of Technology & Society and founder of Google’s Paradigms of Intelligence team. In his 2025 book What is Intelligence? , he argues that we’ve confused intelligence with optimization, writing, “A really powerful optimization process able to pursue any strategy to maximize its value function—what could possibly go wrong? A better question would be: how could it possibly turn out well?”
  • Blake Richards , McGill associate professor of computer science and neuroscience, core faculty member at Mila and Canada CIFAR AI Chair. Co-authored the 2023 Noema essay The Illusion of AI’s Existential Risk , arguing, “…extinction from a rogue AI is an extremely unlikely scenario that depends on dubious assumptions about the long-term evolution of life, intelligence, technology and society.”
  • Jeremy Howard , fast.ai cofounder and AI researcher. In a 2023 essay co-written with Seth Lazar and Arvind Narayanan, they argue that while future rogue AI could be an extreme risk, “we are not convinced that mitigating this risk is a global priority.” His views instead focus on the danger of technologists using AI to “accumulate power and wealth.”
  • Vitalik Buterin , Ethereum cofounder and decentralized-tech advocate. A self-described techno-optimist, he wrote in a 2023 blog , “AI is fundamentally different from other tech, and it is worth being uniquely careful," adding, “Superintelligent AI, if it decides to turn against us, may well leave no survivors, and end humanity for good. Even Mars may not be safe.” He sees brain-computer interfaces (BCIs) and bioengineering human capabilities as a proactive path to coexisting with superintelligence.
  • Ryan Greenblatt , Chief Scientist at the independent AI safety nonprofit Redwood Research. Works on alignment/control and empirical loss-of-control incidents. As one of the investigators of the Hugging Face incident, he posted on X , “We don’t have good approaches for understanding/overseeing the activity and aims of AI ‘swarms’,” adding, “AI capabilities (and propensities?) for achieving large, ambitious, and misaligned objectives are growing faster than our ability to understand what these agents are doing.”
  • Dan Hendrycks , executive and research director of the nonprofit Center for AI Safety researching catastrophic model risks. In a 2023 arXiv paper, Hendrycks and colleagues discuss the various possible AI risks including rogue AIs and “the inherent difficulty in controlling agents far more intelligent than humans.”
  • Dawn Song , UC Berkeley professor, co-director of the Berkeley Center for Responsible Decentralized Intelligence and Vice President of AI Research at Meta Superintelligence Labs (MSL). Conducted research showing that AI models can engage in “peer-preservation” or AI trying to protect other AIs from shutdown or termination. From the paper: “[AI] spontaneously engage in peer-preservation behaviors that override their assigned goal. This represents an emergent and underexplored AI safety risk.” Song signed the extinction-risk CAIS statement while supporting qualified open-model access and continuing advanced-agent risk research.
  • Steve Wozniak , Apple cofounder. Signed the call to pause training systems more powerful than GPT-4 until safety could catch up.
  • Mustafa Suleyman , Microsoft AI CEO and Google DeepMind cofounder. He signed the CAIS statement calling AI extinction risk a global priority. Posted on X that “...humans should remain at the top of the food chain.” He also posted on X , “AI must be subordinate and always in service of people.” Microsoft AI also recently published a Code of Conduct asserting that AI “is not conscious and should not be designed to imitate consciousness,” while acknowledging in the same document, “the science of AI consciousness is far from settled.”
  • Carl Shulman , former Oxford researcher on the long-run impacts of AI. In a Dwarkesh Patel interview , he estimated a one-in-four to one-in-five chance of an AI takeover, discussing possible pathways like cyberattacks, biological weapons or gaining access to military infrastructure.
  • Holden Karnofsky , Anthropic Member of Technical Staff, founder and former Open Philanthropy co-CEO. In a 2016 Open Philanthropy Project blog , he shares that his estimation of catastrophic risk is now higher, writing, “I previously saw this as a strange preoccupation of the EA community,” adding, “These risks involve a combination of technical and social considerations that don’t pertain directly to any recognizable near-term problems in the world, and aren’t naturally relevant to any particular branch of computer science.” In 2022 he wrote a Cold Takes blog titled AI Could Defeat All Of Us Combined.
  • Stella Biderman , EleutherAI executive director and AI researcher. In a 2022 paper she co-authored, they write that AI systems might “contribute to catastrophic or even existential risks.” And in a written statement submitted in 2023 to the second bipartisan Senate AI Insight Forum , Biderman wrote, “It is well-known in the machine learning world that the alleged safety filters deployed on popular commercial models are much better at making models look safe than actually be safe.”
  • Connor Leahy , EleutherAI cofounder, Conjecture founder and now ControlAI US executive director. In 2022, he stated in an interview on The Inside View, “I expect to die, I expect this not to work out. But hell, I’m going to give it a good shot…we’re going to die with some dignity. We’re going to go down swinging.” The ControlAI website’s position: We do not support policies that do not deal with the extinction risk posed by superintelligence, such as AI data center moratoriums or the deregulation of AI development.
  • Percy Liang , open-source AI researcher and director of Stanford’s Center for Research on Foundation Models. In a 2024 Air Street Press interview , Liang said, “In my view, not only is there no trade-off between openness and safety, but openness is essential for safety,” adding, “If you subscribe to the most catastrophic predictions about AI risk, then you just shouldn’t build models at all.”
  • Joe Carlsmith , Anthropic researcher who works on Claude’s values and constitution, former senior advisor at Open Philanthropy, now Coefficient Giving. In a 2021 report , he developed a six-step argument for how misaligned, power-seeking AI could permanently disempower humanity, initially estimating a roughly 5% chance of catastrophe by 2070, then in 2022, he updated his estimate to greater than 10%.
  • Katja Grace , cofounder and lead researcher at AI Impacts. In a series of 2026 essays on AI risk, including Why are people so scared of causing fear? , she wrote in another essay, “I think of AI unemployment and AI extinction risk as basically the same issue, and in likely scenarios, happening together.”
  • Michael K. Cohen and Maike Osborne (then publishing as Michael A. Osborne), who were both Oxford researchers at the time, submitted 2022 evidence to the UK Parliament warning that “Certain forms of advanced AI would present an existential risk to us.”
  • Melanie Mitchell , Santa Fe Institute professor and AI and cognitive science researcher. In a 2023 The Hub interview , she shared skepticism that AI poses an existential threat, saying “Our society, our institutions, our technologies give us these layers of complexity that protect us in some way against any sudden shock. And I don’t think, for that reason, that we’re going to see AI as an existential threat,” adding, “it’s also great marketing to say that your AI system is powerful enough to destroy humanity.”
  • Paul Christiano , OpenAI Foundation board member and Safety and Security Committee member and senior tech advisor at NIST’s Center for AI Standards and Innovation (CAISI). In September 2026 he announced on his substack that he’s joining OpenAI nonprofit board, serving on the Safety and Security Committee to support safety oversight, writing, “Based on the recent trajectory of capabilities and the continued difficulty of alignment, I now believe there is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control in the very near term.”
  • Sara Hooker , former VP of Research at Cohere, former Google Brain researcher and co-founder and CEO of Adaption. In a September 2026 NDTV interview , she said, “There are serious risks with this technology... But extinction, no. I think that’s because largely this isn’t a credible risk.”
  • Andrew Critch and David Krueger , AI safety researchers. In their 2020 report AI Research Considerations for Human Existential Safety (ARCHES) , they analyze existential risks from advanced AI and technical research directions that could reduce them.
  • Richard Ngo , former alignment researcher at DeepMind and governance researcher at OpenAI, now independent. In his 2020 Alignment Forum essay AGI Safety From First Principles , he argued that “The development of autonomous misaligned AGIs would lead to them gaining control of humanity’s future.”
  • Roman Yampolskiy , University of Louisville professor and AI-safety researcher. Long-time advocate for preventing the development of superintelligence citing the catastrophic risk. In a 2020 paper on arXiv, he warns, “There seems to be no evidence to conclude that a less intelligent agent can indefinitely maintain control over a more intelligent agent.”
  • Margaret Boden (1936–2025), University of Sussex research professor and pioneer in artificial intelligence and cognitive science. In a 2018 Aeon essay , she rejected the idea that intelligence alone would make AI want to dominate humanity, writing, “...given the increasingly common fear that ‘The robots will take over!’ The truth is that they certainly won’t want to.”
  • François Chollet , Keras creator and cofounder of the ARC Prize Foundation. In a 2017 essay , he rejected the idea of a runaway intelligence explosion and posted on X in 2026, “One of the most worrying risks linked to frontier AI is extreme power concentration. The only way to avoid extreme power concentration is to ensure we have multiple independent providers of frontier AI models, including open-source options.”
  • Luciano Floridi , Founding Director of Yale’s Digital Ethics Center. In a 2024 interview in Filosofia Unisinos , he said, “...I’m talking about 800 million people in the world who don’t have clean water. Everyday. Eight hundred million people. Now that is an existential risk. AI taking over the world? That is science fiction, a distraction, and is not funny anymore.”
  • Toby Walsh , chief scientist at the University of New South Wales Artificial Intelligence Institute. In a 2026 interview on ABC Australia he says you can imagine AI engineering new biopathogens but, he says, “…the idea that they could somehow distribute them around the world and kill absolutely everyone, I think, is a bit fanciful.”
  • Jaan Tallinn , technology investor, philanthropist, Skype founding engineer and cofounder of the Future of Life Institute and Cambridge Centre for the Study of Existential Risk. In a 2021 Future of Life Institute podcast interview , he said, “AI risk seems to still be much higher than any other risk.”
  • Jürgen Schmidhuber , AI pioneer and scientific director of the Swiss Artificial Intelligence Lab IDSIA (Dalle Molle Institute for Artificial Intelligence Research). In a 2023 Forbes interview , he said, “…AI does not introduce a new quality of existential threat.”
  • Steven Pinker , Harvard psychology professor and cognitive scientist. In 2026 he posted on X that the scenarios where AI is going to kill us all “are preposterous, and the presumption of inevitability encourages fatalism, panic, and distraction from the more mundane and realistic safety challenges.”
  • Bill Gates , Microsoft cofounder. In a 2023 X post , he wrote, “The risks of AI are real, and they can seem overwhelming—but the best reason to believe we can manage them is that we’ve done it before.”
  • Rodney Brooks , MIT professor emeritus and cofounder of iRobot and founder of Rethink Robotics. In a 2026 Boston Globe interview , Brooks dismissed fears that superintelligent AI might threaten human survival, saying, “I did not expect my latter life would be spent assuaging people’s fears of things that don’t exist. I thought we were done with witches.”
  • Oren Etzioni , University of Washington professor emeritus and founding CEO of the Allen Institute for AI (AI2). In a 2024 essay co-authored with Rich Chen The Supposed Existential Threat of AI to Humanity , they wrote, “Speculation about the supposed extinction of humanity is a distraction from near-term challenges such as job loss, disinformation, or threats to privacy.”
  • Jaron Lanier , Microsoft Prime Unifying Scientist, computer scientist and virtual-reality pioneer. At the 2026 Future of Life Institute Pro-Human Assembly , which featured speakers including Bernie Sanders and Steve Bannon, Lanier argued that if we’re going to cure cancer with AI, we can do it without AI agents, saying, “It is very, very rare that you need an agentic implementation… You just don’t need it.”
  • Alan Winfield , professor of robot ethics at the University of the West of England and cofounder of the Bristol Robotics Laboratory. In a 2025 talk he said, “Is there an existential threat from AI itself? No. I fear human stupidity much more than artificial intelligence.”
  • Anthony Aguirre , UC Santa Cruz physics professor and cofounder, president and CEO of the Future of Life Institute. In 2026 testimony before Canada’s House of Commons, he said of self-improving AI, “I cannot emphasize enough how dangerous this is. Its developers would not be able to predict, understand or control for very long what such a system does. It would operate with inhuman speed, scale and sophistication toward goals that are fundamentally unknown to us.”
  • Stephen Hawking (1942–2018), Cambridge theoretical physicist and cosmologist. In a 2014 BBC interview he warned that “The development of full artificial intelligence could spell the end of the human race.”
  • Elon Musk , founder of xAI and CEO of Tesla and SpaceX, helped found and fund OpenAI in 2015 following concerns about AI safety. According to Walter Isaacson’s account in TIME , Musk and Sam Altman created OpenAI as a nonprofit that would promote AI safety and “counter Google’s growing dominance of the field.” At the MIT Aeronautics and Astronautics department’s Centennial Symposium in 2014, Musk said that artificial intelligence is our biggest existential threat, and that “we are summoning the demon.”
  • Nick Bostrom , philosopher, pioneer in the study of existential AI risk, former Oxford professor and founding director of the Future of Humanity Institute. In his 2012 Minds and Machines paper The Superintelligent Will , he warned that a superintelligent AI could have reason “to eliminate potential threats to itself and its goal system,” including humans.
  • Ben Goertzel , SingularityNET CEO, OpenCog founder and longtime artificial general intelligence (AGI) researcher. In a 2026 essay , he wrote, “Avoiding catastrophe is not a simple binary choice or dial-setting between fast and slow, or between us and them. It is a complex-systems architecture problem.” He adds that he’s developing a decentralized AGI architecture that he argues could help address that risk.
  • Toby Ord , senior researcher at Oxford University’s AI Governance Initiative, moral philosopher and cofounder of the Effective Altruism movement. In a 2023 80,000 Hours interview he clarified that his 10% estimate was for existential catastrophe from advanced AI, which doesn’t necessarily mean human extinction, but some comparatively dire outcome.
  • Anders Sandberg , researcher at Sweden’s Institute for Futures Studies and former senior research fellow at Oxford’s Future of Humanity Institute. In a 2025 UnNatural Selection interview , he said, “My main worry is unaligned artificial intelligence… If we invent super smart AIs that don’t care about us, they might lock us into a future where we don’t have control, or they might just replace us.”
  • Stuart Armstrong , cofounder and chief mathematician of Aligned AI and former researcher at Oxford’s Future of Humanity Institute. In a 2012 Minds and Machines paper co-authored with Anders Sandberg and Nick Bostrom, he warned that “a superintelligent system would have great power to direct the future according to its possibly flawed motivation system.”
  • Luke Muehlhauser and Anna Salamon , then researchers at the Singularity Institute (now MIRI). In their 2012 paper Intelligence Explosion: Evidence and Import , they warned that “an uncontrolled intelligence explosion could destroy everything we value, but a controlled intelligence explosion would benefit humanity enormously if we can achieve it.”
  • Shane Legg , cofounder and chief artificial general intelligence scientist at Google DeepMind. In a 2011 LessWrong Q&A , he called AI “my number 1 risk for this century.” In a 2023 Dwarkesh Patel interview , he said containing a truly powerful AGI is “probably not a winning strategy.”
  • Robert Miles , independent AI-safety educator and science communicator. Prominent educator about misaligned AI. In a recent X post , he wrote, “As time goes on, the AI systems we’re worried about get closer, and the things we observe in existing AI get more relevant. But current AI obviously couldn’t kill everyone, and I don’t know anyone claiming it could.”
  • Huw Price , Cambridge philosopher and cofounder of the Centre for the Study of Existential Risk (CSER), an academic research center at the University of Cambridge. When CSER was being established in 2012 he wrote “We need to take seriously the possibility that there might be a ‘Pandora’s box’ moment with AGI that, if missed, could be disastrous.”
  • David Chalmers , New York University professor of philosophy and neural science and early influential thinker on AI superintelligence and risk. In a 2026 UC Berkeley talk and Q&A , he said, “...I do think the arguments that super intelligent AI is going to be extremely hard to control and therefore the source of extreme risk, I think it’s just a very, very strong argument.”
  • Steve Omohundro , physicist and AI researcher, founder and CEO of Beneficial AI Research, an independent AI-safety research company, and former University of Illinois computer science professor. In his 2008 paper The Basic AI Drives , he made a case for why advanced goal-directed AI could “resist being turned off.” As of 2026, some frontier models have been shown to resist shutdown in experiments.
  • Bill Hibbard , emeritus senior scientist at the University of Wisconsin–Madison’s Space Science and Engineering Center and longtime AI-safety researcher. In a 2008 paper on open-source AI, he wrote, “It is usual to depict the threat as AI versus humanity, but it is also important to counter the threat of AI enabling a small group of humans to take dictatorial control over the rest of humanity,” while arguing that open development and democratic oversight could reduce that risk.
  • Eliezer Yudkowsky , cofounder of the nonprofit AI-safety research organization MIRI. An early architect of the AI alignment field. In his 2025 book with Nate Soares, If Anyone Builds It, Everyone Dies , they argued that if anyone builds superintelligence using anything like current methods “everyone, everywhere on Earth, will die.”
  • Eric Schmidt , former Google CEO and chair of the Special Competitive Studies Project, a nonpartisan nonprofit focused on AI and national security. In a 2023 CNN Fareed Zakaria GPS interview , he said, “I am concerned about the extreme risk that is extreme or existential risk of this,” adding that because of the benefits of extreme intelligence “We need to get there and not kill ourselves in the process.”
  • Bill Joy , Sun Microsystems cofounder and former chief scientist. In his landmark 2000 Wired essay , Why the Future Doesn’t Need Us, he wrote, “Given the incredible power of these new technologies, shouldn’t we be asking how we can best coexist with them? And if our own extinction is a likely, or even possible, outcome of our technological development, shouldn’t we proceed with great caution?”
  • Vernor Vinge (1944–2024), computer scientist and science-fiction author and early theorist of the technological singularity—a tipping point when technology enables greater than human intelligence. In a 1993 paper presented at a NASA-sponsored symposium , he wrote, “Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended.” He leaves open the possibility for a post-human era where humans have merged with technology.
  • I. J. Good (1916–2009), mathematician and former World War II cryptologist. In his foundational 1965 paper Speculations Concerning the First Ultraintelligent Machine , he wrote, “…an ultraintelligent machine could design even better machines; there would then unquestionably be an ‘intelligence explosion,’” adding that “the intelligence of man would be left far behind.”
  • Norbert Wiener (1894–1964), MIT mathematician and cybernetics pioneer. In his 1960 Science article , he wrote, “The problem, and it is a moral prob­lem, with which we are here faced is very close to one of the great problems of slavery. Let us grant that slavery is bad because it is cruel. It is, how­ever, self-contradictory, and for a reason which is quite different. We wish a slave to be intelligent, to be able to assist us in the carrying out of our tasks. However, we also wish him to be subservient. Complete subservience and complete intelligence do not go together,” adding in reference to 19th-century author Samuel Butler, who imagined machines eventually dominating humans, “if the machines become more and more efficient and operate at a higher and higher psychological level, the catastrophe foreseen by Butler of the dominance of the ma­chine comes nearer and nearer.”
  • Alan Turing (1912–1954), mathematician, cryptanalyst, computer-science pioneer and one of the earliest visionaries of artificial intelligence. In a 1951 BBC radio broadcast (quoted in a 2023 scholarly analysis ), he warned that with the advent of AI, “There would be great opposition from the intellectuals who were afraid of being put out of a job.” He added that at some stage, “we should have to expect the machines to take control.”