India, May 18 -- Artificial Intelligence (AI) is a technology that is adaptable for many purposes, fed with human linguistic and other behavioural data that is scraped from the immense datasphere existing in the web. Like the forms of software that are necessary for its creation and use, this software, too, is being produced in an unequal competition between parties making "open source" software that anyone can copy, improve and share, and parties making closed, proprietary systems. Over the last few decades, free and open source software (FOSS) has come to dominate the global software industry. Despite the efforts of proprietary-technology companies with enormous resources, sharing rather than excluding turned out to be the way to make better software and, therefore, more money. In the end, even those companies adopted, to varying extent, FOSS licensing and software-making practices. Our lawyering created some of the fundamental licensing and project organisation structures that underlay this immense change. The same process is repeating itself, unsurprisingly, in the layer where machine learning models are at the frontier of software development. Once again, big companies are at the forefront of proprietary development. They have pursued the most capital-intensive strategies for making vertically-integrated intelligent systems. Their goals are to attain "artificial general intelligence" and to give capital an unbeatable bargaining leverage over skilled workers by threatening them with automated replacement. The parties making FOSS AI are intellectual forces across the world's universities, the enthusiastic individuals learning and wishing to learn the technologies, early-adopting US giants, and the immense Chinese industry. The goal is to embed intelligent systems in physical products of every description sold throughout the world. Open source development and sharing gives industry technical flexibility, resistance to anti-competitive oligopoly, and - crucially - design emphasis on lower capital-intensity of production and operation, including radically lower power requirements. Having worked in precisely this corner of law and economics for decades, we know who will win and why. With a small number of staggeringly wealthy multinational firms trying to control the technology, national governments too recognise that the rules of software distribution and the political implications of its operation threaten their autonomy. So "digital sovereignty" is again the motto of the moment, reflecting the anxieties of governments about the potential development of private power so much greater than their own. The phrase seems to evade a precise definition in this environment of hype and alarmist rhetoric. But "digital sovereignty" in this context means what our community has always called "freedom". It means four basic rights: The right to use technology for a desired purpose without requiring additional permissions; the right to study, modify, reuse, and improve the technology; the right to copy, operate at any scale, and the right to share with anyone - all without restrictions, prices, or additional requirements. These freedoms, which underlay our "free software" licensing models, are also what governments want now. Translating those rights into legal guarantees will prevent anti-competitive oligopolies, controlling use of licences and other "intellectual property" limitations, fostering free exchange of knowledge and rapid technical and economic development. This has been the history of the global software industry over the last generation, and it is what governments are urgently seeking now. The US software industry is now built around an effort to achieve goals that are economically incompatible with the welfare of Indian society, with the needs of the Indian government. Chinese technology's technical and political emphases are also unsuited to Indian requirements. India needs intelligent technologies not to build products for export, but to accelerate human development without demanding vast resources or burdening the environment. Those are not the needs of India alone, but of billions of people throughout the Global South and in the social democracies of Europe. This is what India will export in the global AI economy. The path of open source AI is not only the path that our recent history says is most likely to prevail; not only the one that meets the demand for "digital sovereignty", but also one that leads to technology that improves the human future....