The chorus for AI governance is growing stronger, though not everyone is singing the same tune. Senior executives from companies such as Anthropic, OpenAI, and Google DeepMind have been rallying for “pacing the frontier” — an internationally coordinated effort to recalibrate AI’s tempo for better risk management. Meanwhile, civil society organisations would like to see a global moratorium on the development and sale of frontier AI models until binding safeguards are put in place.

On Tuesday, leading US AI companies gathered at the White House to sign a new Accord on Super Intelligence, which takes a considerably softer, voluntary approach. The “morally binding” accord speaks of building robust internal controls, independent external evaluation, and oversight by an independent committee of the company’s board. However, the actual standards and enforcement mechanisms are not specified, affording significant discretion to the participating companies.

Advertisement

This comes in response to multiple incidents of unauthorised and deceptive behaviour exhibited by frontier AI agents. AI-related risks were also prominently discussed during the recent UN General Assembly. Finland and Norway initiated an international effort, now endorsed by 28 countries, through “A Call for Control of Frontier AI Models”. The signatories are seeking coordinated international action on the governance of frontier AI. Their three-step agenda includes mandatory pre-deployment testing and independent evaluation for AI companies, common incident reporting standards, and an institutional mechanism to enable standard-setting and verification. These are significant proposals, but their path to implementation remains unclear.

Notably, neither the US nor China, the two nations at the forefront of the geopolitical dynamics of frontier AI, joined the international call. The US continues to emphasise permissionless innovation, with expectations of responsible conduct by AI companies as its preferred approach. It has also been emphatic in rejecting any “globalist scheme of control for superintelligence” at the UN level or otherwise. China, in contrast, has maintained that it respects the UN’s role as the main channel for global AI governance but remains wary of what it labels as Western-centric agendas.

In July this year, Beijing initiated an alternative institutional path for global AI governance through the World Artificial Intelligence Cooperation Organisation (WAICO). Twenty-nine countries became founding members of the Shanghai-headquartered organisation. Its functions will include promoting “supply-demand matching” of AI, standard setting, and international convergence on global AI governance.

Advertisement

WAICO’s membership is open to any interested country. Some commentators have contrasted this with the liberal-democratic and rule-of-law assumptions associated with other AI initiatives, such as the Global Partnership on Artificial Intelligence and the G7 Hiroshima Process. As an intergovernmental organisation, WAICO could also serve as a site for formal consensus-building that goes beyond the soft-law-centric approaches in other forums. Yet, its positioning as a China-led initiative could limit the practical expansion of WAICO’s footprint outside the Global South.

In a positive development, the US and China recently opened the door for a bilateral dialogue to debate the risks and benefits of advanced AI. But a broader consensus on AI governance remains elusive. That, however, cannot be a reason for inaction by the rest of the world. Neither should it become an excuse to allow the tech industry to play an outsized role in defining what constitutes a suitable mode of governance. However, the latest White House accord places the ball firmly in the industry’s court.

Given these realities, where should India locate itself in the global AI governance debate? As a country that is admittedly not at the forefront of frontier AI but is seeing extensive deployment of such models, India has a direct interest in risk-mitigation and accountability safeguards for frontier AI. Its positioning on digital sovereignty also supports a claim to greater representation and agency in defining the rules of the game. This serves as a rational basis for India to lend its support to the Finland-Norway proposal.

Finally, the hype around frontier AI should not drown out the more immediate needs of robust AI governance in the country. Rather than waiting for the catastrophic risks and possibilities of artificial general intelligence that may arise in the future, we need impactful legal and institutional safeguards to govern the AI systems that are already embedded in everyday life.

The writer is a lawyer and digital rights researcher