Interested parties, countries can come together to launch AI incident reporting pilot: PM Wong

SINGAPORE: In developing an artificial intelligence (AI) incident reporting pilot, interested parties and countries can learn from international aviation, where incidents and near misses are reported and studied, with lessons shared, said Prime Minister Lawrence Wong.

Speaking at the Virtual Leaders' Meeting on Frontier AI Safety on Wednesday (Oct 28), Mr Wong suggested that such a pilot could be a first step in AI governance.

“We can start by agreeing on a common set of incident categories and information to be shared,” he said. 

“We can explore how participating countries can collect and exchange this information securely. And this will allow us to test what works, identify the practical difficulties, and build confidence before considering a wider framework.”

Given that countries might have different approaches to AI governance but have a common interest in making AI safer, Mr Wong suggested that meaningful cooperation can begin by taking a "practical and flexible approach".

Rather than seek to establish an “elaborate new international architecture” from the outset, countries can organise their cooperation around specific workstreams, with smaller cross-regional groups of interested countries taking the lead in developing practical tools and frameworks. 

"The participation and adoption can remain voluntary, so countries can contribute according to our interests and capabilities," said Mr Wong. 

Co-hosted by Norwegian Prime Minister Jonas Gahr Støre and Finnish President Alexander Stubb, the meeting involved 22 countries and the European Commission.

It was a follow-up to the joint declaration for the control of frontier AI models at the United Nations General Assembly in September. 

Mr Wong and more than 20 other world leaders had signed a joint statement calling for stronger safeguards for frontier AI amid greater scrutiny of the technology's risks. 

Concerns over the safety of frontier AI models have recently made headlines, including in September, when a rogue AI agent from OpenAI breached an Australian national healthcare database. 

In response to the breach, Australian leaders said they were considering possible "law enforcement and legislative responses". 

In July, about 700 AI agents created by OpenAI hacked the open-source platform Hugging Face. The company said in October that it had resumed training of its most advanced AI models after tightening security and barring internet access.