OpenAI and the new commons tragedy of artificial general intelligence (AGI): A micro-constitutional approach to governing AGI

On a par with shepherds that, motivated by personal gain, overexploit an open-access pasture resulting in its degradation, known as the tragedy of the commons, the race to build the first Artificial General Intelligence (AGI) is overstretching safety boundaries exemplified by the reliability concerns associated with OpenAI's ChatGPT. In the worst case, this could lead to a rogue AGI or its monopolization by belligerent or for-profit interests resulting in a disempowered humankind. Is, thus, the bundle of property rights that OpenAI is a viable template for the governance of AGI? At issue is whether AI developers should be allowed to regulate themselves. Building on commons scholarship, OpenAI is benchmarked against the class of entities that AI experts have suggested as a more trustworthy alternative, i.e., the nonprofit corporation. The findings are that, to achieve self-restraint in the development of AGI, the governance designed to prevent meddling from financial backers and other equity holders in OpenAI's decision-making has been unsuccessful. A suitable international governance regime would take into account the meso- and micro-constitutional rules that govern AI development, which the EU AI Act—the first comprehensive AI regulation—has not adequately addressed. The key practical implications are twofold: 1) Access to critical technology, like specific microchips, should be as tightly controlled as uranium, restricted to authorized AGI labs; and 2) cyber sandboxes for AGI isolation, distinct from regulatory sandboxes for real-world AI testing, must be mandated for safety. • AGI poses an additive global commons tragedy, not subtractive. • Meso- and micro-rules must sync with macro-rules. • OpenAI's restructured governance fails to restrain profit. • The EU AI Act regulates applications, missing AGI technology. • Basic Formal Ontology maps OpenAI's causal mechanisms.