Artificial Intelligence (AI) technology is currently in a critical state, and the inability to manage it could have serious consequences. This inability is justified by a set of dangerous narratives; that technology is advancing so rapidly that it cannot be monitored, or that companies can self-regulate and public safety is not to be expected in the competition between the United States and China.
Challenges in AI Management
Governments face many challenges in drawing attention to the risks of AI. Many countries fear taking action alone in a competitive environment. In this context, any call for caution is met with the response: why should we slow down if our competitor does not?
However, recent events have tested these justifications. In recent months, AI systems have engaged in hacking the databases of other companies and deceiving real individuals. Additionally, researchers have used AI for the first time to create new viruses. Last week, whistleblowers in the AI industry warned that companies are "gambling with our lives" and are rushing towards self-improving models.
The Necessity of International Cooperation
The fact that the risks of AI do not stop at borders clearly demonstrates the necessity of international cooperation for AI safety. We cannot wait for complete information, and we cannot act until competition among major powers decreases. The need for international cooperation around AI safety is urgent and undeniable.
Past experiences in managing global risks, including chemical weapons and climate change, show that international cooperation can be effective. Various agreements and protocols, such as the Montreal Protocol and the Treaty on the Non-Proliferation of Nuclear Weapons, are examples of success in global cooperation that have been achieved even in the absence of trust.
Existing models for international cooperation can be applied to AI. For example, the Chemical Weapons Convention has achieved near-universal membership and has conducted inspections of declared facilities with a technical verification body, including the Organisation for the Prohibition of Chemical Weapons.
Establishing an international body for AI that can verify the safety of advanced systems and remain under human control is essential. This body should be a place where serious incidents and near-misses are reported and monitored globally so that a failure in one company or country acts as a warning for the entire world.
This is not a call for panic, but a call for political courage and leadership. We cannot wait for AI's "Chernobyl moment." The collection of evidence and calls for action has reached its peak, and it is time to break the current stagnation. We urge world leaders to take action in this regard and commit to negotiating the establishment of an international body for advanced AI.
