The co-founder of Anthropic, Dario Amodei, has issued a stark warning: artificial intelligence risks evolving without meaningful human oversight unless nations like Britain step up to define the ethical boundaries of this technology. Speaking at the Royal Society in London, Amodei argued that the current trajectory of AI development prioritises speed over safety, leaving humanity vulnerable to systems that may act in ways we cannot predict or control.
Amodei’s comments come as the UK government prepares for its AI Safety Summit, a global gathering aimed at establishing norms for the burgeoning industry. He praised Britain’s proactive stance but cautioned that rhetoric must translate into binding commitments. “Without a shared framework, we risk a race to the bottom where convenience trumps consequence,” he said.
The core of Amodei’s argument rests on the concept of alignment: ensuring AI systems pursue goals that align with human values. He pointed to recent incidents where large language models exhibited unexpected behaviours, from generating misinformation to bypassing safety guardrails. “We are building intelligence without wisdom,” he warned. “It is like giving a child a weapon without teaching it morality.”
Britain, with its deep expertise in ethics and law, is uniquely positioned to lead. Amodei cited the country’s history of regulating disruptive technologies, from the Industrial Revolution to the internet. “The UK can be the conscience of the digital age,” he said. He urged the government to invest in AI safety research and create a regulatory body with real enforcement power, akin to the Human Fertilisation and Embryology Authority.
But the challenge is immense. AI development is largely driven by private companies accountable to shareholders, not society. Global cooperation is necessary but fraught with geopolitical tensions. Amodei acknowledged these hurdles, insisting that Britain must act as a bridge between the US and China, articulating a third path that values human dignity over economic gain.
The audience, a mix of policymakers and tech executives, responded with cautious optimism. A senior Whitehall official confirmed that the government is considering a “digital sovereignty” framework that ensures British citizens’ data and values are prioritised in AI systems. The summit’s agenda includes sessions on transparency, bias, and the existential risks of general artificial intelligence.
Amodei concluded with a vision: a future where AI amplifies human potential without replacing it. “We need not fear the machine,” he said. “We need to ensure it remains a tool, not a master.” Britain, he believes, can write the rulebook for this relationship. The question is whether the world will follow.
