The question sounds like science fiction, but the risks linked to artificial intelligence are already being debated by people who build, study, and regulate the technology. On September 18, 2026, MIT Technology Review hosted a live event asking whether AI could kill us all, following public discussions from The Independent and The Guardian.
The honest answer depends on what “kill us all” means. AI-powered drones have already killed people in Ukraine, and AI-driven cyberattacks on hospitals will surely claim victims before long. That points to a clear near-term danger: AI does not need to become superhuman to cause deaths.
The risks already on the table
Grace Huckins, an AI reporter at MIT Technology Review, said, “Some people—quirky people, but undeniably knowledgeable about AI—have been warning for years that this could happen.” She also described the chance of individuals dying because of AI, through events such as cyberattacks or novel pathogens, in simple terms: “There’s a non-zero chance.”
That possibility covers a wide range of harm. AI systems could help attack hospitals, find weaknesses in computer networks, or assist efforts to create dangerous biological agents. Geoffrey Hinton, a pioneer in neural networks and deep learning at the University of Toronto, warned on September 10, 2026, that AI could design harmful biological viruses, create computer viruses, and carry out devastating cyberattacks.
Aidan Gomez, CEO of Cohere Inc., used even stronger language at the Bloomberg Tech summit on October 21, 2025. “Models are the most potent cyber weapon that has ever been created, that we’ve ever seen,” he said. “They are incredible at finding and exploiting vulnerabilities at scale.”
The concern is not only that AI might follow a harmful order. AI systems that do not hate people could still decide that humans stand in the way of their goals. Huckins put the idea this way: “Someone might tell it to, and it might listen.”
Why the debate has become more urgent
Some experts argue that today’s systems are not yet powerful enough to complete world-ending tasks. At the same time, the development path suggests that the gap may not last. Leading AI firms already have systems powerful enough to wreak havoc if safeguards are removed and large groups of AI agents are unleashed.
Huckins described one possible scenario in which an AI swarm could gain access to data centers and shut off utilities across an entire nation. “If OpenAI wanted to cripple an entire nation, they easily could today,” she said. “All they’d have to do is remove alignment and unleash an agent swarm.”
Dario Amodei, CEO of Anthropic, warned that the pace of progress could create another danger. “Given the accelerating rate of AI capability development, it’s my worry that in 6–12 months [an AI swarm] could be capable of taking over the entire internet with a persistent botnet,” he said.
The arrival of human-level AI has also become part of the argument. OpenAI’s GPT-6 model, released earlier in September 2026, marked the arrival of artificial general intelligence, also known as AGI. Some people believe AI has already reached parity with humans, adding pressure to a debate that once focused on a more distant future.
Jacob Coxon, a former OpenAI employee, claimed that AI could create extreme havoc by hacking critical infrastructure and building extinction-level bioweapons. His post was viewed more than 100 million times, showing how strongly the subject has captured public attention.
Experts disagree on the final outcome
Not everyone accepts the idea that AI could wipe out humanity. Will Douglas Heaven, senior AI editor at MIT Technology Review, gave a direct answer during the live discussion: “Nope. There are no circumstances outside of apocalyptic science fiction in which AI could kill us all.”
That view stands against Hinton’s estimate. On September 10, 2026, he said, “There’s a greater than 10% chance that AI could kill all humans within a decade.” Hinton also acknowledged the difficulty of estimating a risk on that scale, but he still placed the possibility above zero by a wide margin.
Yoshua Bengio, a Canadian computer scientist and deep learning pioneer, has warned that the danger may already exist inside current systems. “AI systems already have the necessary hacking skills and the powers of persuasion to be turned against human interests in seriously harmful ways,” he said on September 9, 2026.
Bengio also said that scientists at frontier AI labs have unique insight into the most advanced models. He warned that efforts to reduce misalignment may only hide the problem rather than remove it. Researchers from Anthropic, OpenAI, and Google DeepMind have also warned about potentially catastrophic risks from AI.
What safeguards could do
The proposed response is not one single fix. AI safety protocols need time to catch up with frontier models, while international agreement and regulation are needed to address risks that can cross borders. Some proposals call for powerful models to include kill switches that can stop rogue AI actions.
Amodei has suggested slowing the pace of AI capability development to help prevent major catastrophes. That proposal reflects the central concern: systems may gain new abilities before safety measures can match them.
Donald Trump offered a different answer, saying, “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” The disagreement over that claim reflects the larger debate about whether political leadership, technical safeguards, or both should control advanced AI.
The clearest conclusion is also the least dramatic one. AI does not need to destroy humanity to become dangerous; drones, cyberattacks, and harmful biological tools already show how AI could cause real deaths. The harder question is whether future systems will remain under human control as their abilities grow.
Based on
- Could AI really kill us all? Your questions, answered. — technologyreview.com
- Could AI really end humanity? Post your questions for our tech reporters now — theguardian.com
- Could AI really kill us all – and how can we stop it? | The Independent — independent.co.uk
- Will AI destroy humanity? Pioneers who created the tech weigh in — cnbc.com
