Bill Gates has warned that artificial intelligence could lead to a billion deaths unless governments, technology companies and civil society organisations work together to prevent the most dangerous outcomes. The Microsoft co-founder said no single group can be trusted to manage the risks alone, and that self-regulation by the companies building the most powerful systems is not sufficient.
«No one thinks self-regulation is enough,» Gates said, according to remarks reported by Inc. His intervention places one of the world's most prominent technology figures at the centre of a growing debate over who should set the rules for advanced AI, at a time when governments in Britain, the European Union and the United States are still drafting their approaches to the technology.
The warning reflects a broader shift in the AI safety conversation. Until recently, much of the public argument focused on long-term, speculative risks. Gates's framing is more immediate: the danger is not only that machines might one day act against human interests, but that the systems are being deployed faster than the institutions meant to oversee them can respond. In that gap, he argues, lies the potential for catastrophic harm.
His call for coordination identifies three broad groups that must be involved. The first is governments, which hold the legal authority to set safety standards, require testing and impose penalties. The second is the technology industry itself, which possesses the technical knowledge and the ability to build safeguards into models before they are released. The third is civil society, including researchers, public health bodies and advocacy groups that can scrutinise claims, represent affected communities and hold both governments and companies to account.
The argument is that each group has a distinct role and none can substitute for the others. Companies may understand the technology best, but they face commercial incentives that can conflict with caution. Governments can legislate, but often lack technical expertise and can be slow to act. Civil society can raise alarms, but has limited power to enforce change. Only a framework that combines all three, Gates suggests, can reduce the risk of the most severe outcomes.
For Britain, the debate is particularly relevant. The country has positioned itself as a centre for AI research and investment while also hosting international discussions on AI safety. British policymakers have favoured a lighter-touch, pro-innovation approach, but pressure is growing for clearer statutory rules, especially as AI systems are adopted in healthcare, finance, defence and public services. Gates's warning adds weight to arguments that voluntary commitments alone will not be enough.
The reference to a billion deaths is deliberately stark, and it is likely to draw criticism from those who see such figures as alarmist. But it also serves a purpose: it forces the question of what level of risk is acceptable when the technology is general-purpose and increasingly embedded in critical systems. Even a small probability of a catastrophic outcome, if the systems are widely deployed, becomes a serious concern for insurers, regulators and investors.
Businesses are already grappling with the practical side of this. Companies adopting AI face questions about liability, data security and compliance, and many are waiting for clearer rules before making large commitments. A coordinated international approach would reduce uncertainty, but the current landscape is fragmented, with different jurisdictions moving at different speeds and in different directions.
Gates's intervention is unlikely to settle the argument. It does, however, sharpen it. The choice, as he frames it, is not between innovation and safety but between coordination and drift. Without cooperation among governments, industry and civil society, the risks will be managed unevenly, if at all. With it, there is at least a chance of preventing the worst outcomes before they occur.