Preventing corporate capture requires shifting focus from voluntary guidelines to enforceable, binding standards. Currently, big tech firms set the pace because they own the compute and the datasets. Regulators must move beyond mere advice. They can achieve this through mandatory transparency requirements regarding training data and algorithmic decision-making processes. When companies must reveal the ingredients of their models, governments regain the ability to audit them for bias or safety risks.
Antitrust laws also play a central role. If a few firms control the essential infrastructure—like high-end chips and massive cloud networks—they effectively govern the digital world. Breaking up these monopolies or imposing heavy interoperability requirements ensures that smaller players and public institutions can compete. This prevents a single vendor from becoming the gatekeeper of intelligence itself.
Public investment in sovereign AI infrastructure offers a direct counterweight. States should fund national research labs and open-source models that do not answer to shareholders. By building public-interest technology, governments ensure that the trajectory of AI serves societal needs rather than just quarterly earnings. Power follows where the resources lie; moving resources toward the public sector is the most direct way to rebalance the scales.
