Last month, China show di world dia plan to create one global body wey go dey manage artificial intelligence (AI). Di idea na to make sure say everybody fit get equal access to di technology, encourage beta collaboration between countries, and solve wahala like hardware restrictions and di limit wey dey for talent exchange.

Six months before dat one, 58 countries gather for Paris to support one declaration wey dem call “inclusive and sustainable AI.” But di United States and United Kingdom no gree sign am. Even though di approach dey different, di pattern still remain di same. Di race to control AI dey move faster pass di effort to regulate am.

Artificial intelligence na big tool wey dey boost economy, security systems, and geopolitical power. Over di past ten years, e don waka comot from tech labs enter military targeting systems, government procurement, and di main infrastructure wey countries dey use. Di question now na whether e fit dey properly regulated or not.

When di same AI model fit help person pass medical licensing exam today, and tomorrow e fit create realistic deepfake videos, e don clear say di tools wey dey ground to monitor AI no fit match di speed and power wey di technology get.

Di regulatory systems wey dey ground before na for slow advances and small applications. Dem no design am for systems wey fit produce human-level text, analyse satellite images in seconds, or control fleets of autonomous machines. Dis gap na wetin dey make people dey call for new ways to manage AI wey go fit handle di speed and di kind wahala wey e fit cause.

For international level, di way wey new technology dey change di risk for conflict and instability na wetin dey drive di governance debate. Some technology fit reduce risk by improving communication or verification, but others, especially di ones wey dey give advantage for battlefield, dey increase military spending and arms race. AI dey fall for di second group.

Di success of AI for surveillance, targeting, and autonomous systems dey make people believe say if dem no catch up, dem go dey at serious disadvantage. In theory, agreements fit help set limits to avoid competition wey no dey stable. But for reality, dis kain deals no dey common because of distrust and di dual-use nature of AI.

For inside countries, regulation fit work if di government get di power and knowledge to manage am. But for global level, di way AI dey spread make am almost impossible to create rules wey everybody go follow.

For domestic level, governments fit pass laws, create rules, and make sure say developers dey comply. Dis na where culture, law, and economy fit reflect for di way dem dey manage AI. But di ability to regulate AI no dey equal for all countries. E need advanced infrastructure, people wey sabi di technology, and institutions wey fit adjust rules as di technology dey change.

Di speed of AI na one big challenge. AI fit change within months, but lawmaking dey take years. Countries wey go fit keep up na di ones wey go use flexible methods like rolling standards or regulatory sandboxes to update rules quickly.

Another challenge na expertise. Many regulators and civil servants no get di technical knowledge to understand di risks or di systems. Without dis knowledge, di rules wey dem go make fit no work well. Dis na serious problem especially for critical areas like healthcare or financial systems.

Dependence on foreign AI platforms na another issue. If country dey depend on AI systems wey dey trained or hosted abroad, e go hard for dem to enforce dia own standards. Without independent capability, regulators no go fit do much.

For international level, di matter dey more serious. AI dey promise big gains for military, and no big power wan slow down. Di fear of falling behind dey make restraint hard. Di global military AI market dey grow fast, and di money wey countries dey spend show say dem no plan to stop anytime soon.

Even if governments wan slow down, di problem of verification go still dey. Unlike nuclear weapons wey easy to count and monitor, AI models fit dey trained in secret and stored anywhere. Dis make am hard to trust any agreement.

Di dual-use nature of AI dey complicate things. Di same technology wey fit help for healthcare fit also work for military purposes. Dis make treaties wey go separate military AI from civilian AI very difficult.

Di state of di international system no dey help. Trust among big powers dey low, and multilateral diplomacy dey struggle. Discussions for di UN about killer robots don dey drag for years without solution.

Di likely result na fragmented international environment. Countries go dey pursue AI weapons and intelligence systems without agreed rules. Di best way forward na for governments to build capacity at home and focus on small, enforceable agreements for di most dangerous areas internationally.