UK Greens propose global AI regulator and ban on autonomous lethal weapons

A five-part political challenge

Britain’s Green Party has proposed a five-point programme to curb the power and security risks of advanced artificial intelligence. Sky News reported on September 10 that the package includes international cooperation to establish a global AI oversight body and a legal prohibition on lethal autonomous weapons. The proposal marks a political intervention rather than a change in UK law, but it places military autonomy and corporate power inside the same regulatory debate.

The party’s case is that frontier systems are developing faster than national institutions can supervise them. Its plan seeks stronger public control over large technology companies alongside international rules for systems capable of causing severe harm. A worldwide regulator would require negotiations with other governments, while a weapons ban would have to define which systems exercise unacceptable autonomy and what degree of human control must remain.

The weapons component exposes a direct policy fault line. In June, the UK government announced more than £5 billion for drones and autonomous weapons as part of its Defence Investment Plan. The programme envisages uncrewed naval systems, autonomous aircraft working alongside crewed fighters and expanded battlefield-drone capabilities. The government describes those investments as a way to improve deterrence and military effectiveness.

The unresolved question of human control

The United Nations has identified autonomous weapons capable of attacking civilians without human control as a threat to the protection of civilians. Secretary-General António Guterres has repeatedly argued that machines must not make life-or-death decisions without meaningful human authority. That official position supports the international relevance of the Greens’ proposal, although it does not endorse the party’s full domestic platform.

Existing UK policy says the country does not possess fully autonomous weapon systems and has no intention of developing weapons that select and attack targets without context-appropriate human involvement. At the same time, the government argues that some defensive systems may need to act faster than direct real-time supervision permits. The disagreement is therefore not simply whether humans matter, but how human control should be defined, audited and enforced as operational speed increases.

The next stage is political rather than technological. The Greens would need parliamentary support for national restrictions and diplomatic partners for a global institution or binding weapons agreement. The government will face pressure to explain how its expanding autonomous-defence programme fits its existing ethical commitments. Until legislation, a treaty or an executive decision follows, the five-point package remains a proposal and does not qualify as a military alert.