The EU’s Digital Services Act can impose fines of up to 6% of global annual turnover and, in extreme cases, seek a temporary service suspension. Global frameworks such as the OECD AI Principles remain non-binding, leaving states to decide whether cross-border technology risks require common enforcement or flexible national approaches.
Sources
Technology supply chains ignore national borders, but victims and regulators are stuck inside them. The OECD AI Principles are explicitly non-binding, so adherence can become branding. A treaty with minimum standards would stop companies shopping for the weakest jurisdiction while still allowing countries to adopt stronger protections.
Voluntary principles work best for actors already inclined to behave. The DSA can impose fines up to 6% of global turnover and daily penalties for continued noncompliance. That gives a rule weight across borders. Start with narrow duties such as incident reporting, researcher access, and child-safety risk assessments, then let independent courts review enforcement.
Voluntary principles have a jurisdiction problem: a company can simply route around them. When one country tightens AI or data rules, development shifts to wherever the rules are loosest, and users everywhere live with the consequences. That is regulatory arbitrage, and it is exactly what happened with data havens and tax havens. The OECD AI Principles are explicitly non-binding, so compliance is a branding choice. Enforceable international rules set a floor nobody can undercut, the way aviation safety standards work through the ICAO. Without a floor, the most responsible companies face the highest costs, which punishes exactly the behavior we want. A binding minimum does not require a world government. It requires countries to agree that some harms should be illegal everywhere.
There is no neutral global regulator waiting to take the job. Governments disagree on speech, privacy, surveillance, and due process. An enforceable body could give authoritarian states leverage over tools used elsewhere. The UN AI proposals focus on science, standards, capacity, and coordination for a reason: consensus is more realistic there than on binding sanctions.
The DSA shows enforcement is possible within a political and legal union, not across the whole world. Its 6% fine and suspension powers rest on EU courts and institutions. International rules should stay interoperable and voluntary while national regulators enforce their own laws, then cooperate on evidence and cross-border cases.
Treaties take a decade; technology takes eighteen months. The DSA was proposed in 2020 and its full obligations only bit in 2024, an eternity in AI years. European Commission. A binding global tech treaty would be negotiated by diplomats, ratified by 190 parliaments, and arrive just in time to regulate the previous generation's problems. Worse, it would lock in those answers: once a standard is international law, updating it requires another round of global consensus, which means the rules ossify while the technology sprints. Voluntary frameworks can be revised in a year. National regulators can move in months. A binding treaty is the slowest possible instrument aimed at the fastest-moving target, and its main achievement would be giving everyone the comforting illusion that the problem was handled.
Who do you vote out when the global tech regulator gets it wrong? Binding international rules move decisions from elected parliaments to treaty bodies and expert committees, and there is no ballot box for those. A national regulator that overreaches can be hauled before a legislature, sued in domestic courts, or reversed after an election. A global enforcement body answers to no voters, which is fine for technical standards like aviation safety and dangerous for questions as political as speech, surveillance, and AI. The DSA works because the EU has a parliament, courts, and democratic legitimacy behind its fines. EDAA. A global equivalent would have the fines without the legitimacy. When enforcement is detached from democratic accountability, the rules serve whoever captured the process. Keep the power where the voters are.
A binding global rule gets written by committee, which means it gets written to the lowest common denominator. The result locks in weak standards and, worse, can preempt stronger national laws: once a global ceiling exists, industry lobbies to treat it as the final word. Voluntary frameworks let leaders go further, and history shows norms often harden into law after the best practices are proven. The EU's DSA can levy 6 percent fines because it sits inside a real legal union with courts; a global equivalent has no such foundation. Better to let ambitious jurisdictions compete upward than to chain everyone to a compromise drafted to satisfy 190 governments.