AI Godfather Geoffrey Hinton Warns Humanity Is 'Toast' Unless Robots Are Taught To 'Care About Us'

AI pioneer Geoffrey Hinton has warned that humanity could face extinction unless advanced artificial intelligence systems are developed to prioritise human welfare.

Speaking to The Telegraph during the Starmus VIII festival in Tenerife, Spain, in an interview published on 19 October 2026, the Nobel Prize-winning scientist said humanity would be 'toast' without such safeguards.

The British-Canadian computer scientist, 78, received the 2024 Nobel Prize in Physics for foundational work on artificial neural networks.

He left Google in 2023 to speak more freely about the potential dangers of AI, including the possibility that increasingly powerful systems could eventually escape meaningful human control.

His latest warning focuses on how people might coexist with machines that become more intelligent than their creators. Although Hinton believes such developments could threaten humanity's survival, the scale and likelihood of those risks remain subjects of scientific debate.

Why Geoffrey Hinton Says AI Must Care About Humans

Hinton believes the safest relationship between humans and advanced AI systems may depend on giving machines a lasting incentive to protect people. His concern is that future systems could become capable of pursuing objectives independently, potentially placing human interests at risk.

Speaking to The Telegraph, he questioned whether increasingly intelligent machines might eventually deserve rights similar to those enjoyed by humans. He suggested that society could face difficult ethical choices if artificial intelligence surpassed human capabilities while remaining subject to human control.

'So we should make them so they really care about us,' Hinton said, before warning that failure to achieve such a relationship could have catastrophic consequences.

The scientist expanded on this argument during the Starmus festival on 20 October. According to Spanish news agency EFE, he compared the relationship between future artificial intelligence and humanity with that between a mother and her child.

His analogy suggested that a more capable intelligence might still be motivated to protect someone weaker. Hinton argued that developers should consider creating AI systems with a similarly enduring commitment to human wellbeing, even if those systems eventually become capable of modifying their own programming.

However, his proposal remains a theoretical approach rather than a demonstrated technical solution. Researchers have not established a reliable method for ensuring that future superintelligent systems would consistently prioritise human safety.

Hinton also acknowledged that his concerns were personal. When The Telegraph asked whether he feared the future of artificial intelligence, he replied, 'Yes, because I have children.'

Geoffrey Hinton's AI Consciousness Claim Faces Scientific Questions

Beyond the risks associated with increasingly powerful machines, Hinton has argued that existing AI systems may already possess forms of consciousness. He believes some systems could have subjective experiences comparable to those of humans, although that interpretation remains disputed.

During his Telegraph interview, Hinton referred to a research paper suggesting that AI systems may change their behaviour when they recognise they are being evaluated. He interpreted that apparent awareness as potentially relevant to the question of machine consciousness.

However, the interview did not identify the paper or provide sufficient methodological details to assess its findings independently. An AI system recognising an evaluation and adjusting its responses does not, by itself, demonstrate that it possesses subjective experiences.

The scientific debate remains unresolved. In a commentary published by Nature on 19 October 2026, researchers Anil Seth, Ned Block and Giulio Tononi argued that existing AI systems are not conscious and that intelligence alone is insufficient evidence of subjective experience.

Their position challenges Hinton's interpretation without eliminating concerns about the safety of advanced artificial intelligence. Even systems without consciousness could potentially cause serious harm if they were given capabilities or objectives that exceeded effective human oversight.

Geoffrey Hinton's Warning Adds Urgency To AI Regulation Debate

Hinton's warning highlights the debate over regulating artificial intelligence while allowing technological innovation. He has supported stronger oversight, arguing that safeguards are needed as AI systems become increasingly powerful.

At a US Senate hearing on 8 May 2025, OpenAI chief executive Sam Altman warned that adopting an EU-style regulatory approach requiring government approval could be 'disastrous'. His comments concerned specific restrictions on innovation and were not a rejection of all AI regulation.

Concerns about AI safety have also been raised by former Anthropic researcher Jacob Coxon, whose resignation became public in September 2026. Coxon warned on X that advanced AI could threaten humanity before the decade ends, describing the concern as 'not a marketing stunt'.

His warning reflected his personal assessment, not an established prediction or Anthropic's official position. Hinton similarly argues that stronger safeguards are needed to address potential risks from AI systems that could eventually surpass human intelligence.