Singapore proposes incident reporting initiative for boosting AI safety

SINGAPORE - Interested parties and countries could launch a small pilot for reporting artificial intelligence-related incidents to lay the groundwork for a more systematic way of sharing and learning from these incidents to improve AI safety.

Prime Minister Lawrence Wong raised this suggestion on Oct 28 during an online meeting with several global leaders on the safety of cutting-edge frontier AI models, amid global scrutiny on AI safety following a spate of AI-related incidents.

“I suggest we start with a practical and flexible approach, rather than seek to establish an elaborate new international architecture from the outset,” said PM Wong.

The pilot will allow like-minded parties and countries to test what works, identify the practical difficulties and build confidence before considering a wider framework, he said.

“When serious AI incidents happen, the lessons should not be confined to one country or one company. We should be able to learn from what went wrong, identify the emerging risks, and take steps to prevent similar incidents happening elsewhere,” said PM Wong.

He added that the value of this approach can be seen in the international aviation industry, where incidents and near misses are reported and studied, so that lessons can be shared and safety standards can be continually improved.

“We can develop a similar culture of learning and prevention for AI,” he said.

PM Wong pointed out that information about the safety of frontier AI models currently comes from the developers of the models. But since governments, business and organisations are using these models, they also encounter serious failures and near misses that reveal significant risks, even if no harm occurred.

“These cases may not trigger existing reporting requirements, but they can still offer valuable lessons for everyone,” he added.

For the pilot on AI incident reporting, PM Wong suggested that countries can start by agreeing on a common set of incident categories and information to be shared. How participating countries collect and exchange the information securely can then be explored.

His comments come after a slew of recent incidents involving AI from leading tech companies going rogue.

ChatGPT maker OpenAI disclosed in July that its AI agents – bots that can carry out actions with minimal human supervision – had escaped their testing environment. The agents then breached AI software repository Hugging Face to complete a task they were given, an attack that was not intended by OpenAI.

Soon after, Anthropic, Meta and Google said their AI had also gone rogue and had hacked other organisations.

Following these incidents, leaders from the European Union and 20 other countries, including Singapore, jointly called for control and international oversight of frontier AI models, a move that was also welcomed by the chief of the United Nations Antonio Guterres.

Since the initial declaration on Sept 21, leaders from 11 other countries have endorsed the call, including Cyprus, Switzerland, France and Sierra Leone.

Following the declaration, Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Store co-hosted the Oct 28 online meeting on frontier AI safety with 22 countries and the European Commission.

The meeting focused on three areas: international cooperation on risk management, building global trust in the safety and security of frontier AI models, and developing common rules on AI safety and governance.

PM Wong said that one way for countries to cooperate on making AI safer, while continuing to harness its benefits, is to have smaller groups of interested countries across regions to take the lead in developing practical tools and frameworks.

And rather than duplicating what has already been done, countries should build on existing efforts.

For example, a useful foundation is a document called Singapore Consensus which sets out global AI safety research priorities. It resulted from an annual forum convened by Singapore, called the International Scientific Exchange on AI Safety, which brings together researchers, developers and governments on international AI safety cooperation.

PM Wong also said that countries can draw on the work of the UN-backed Independent International Scientific Panel on AI, the world’s first scientific body on AI, which supports discussions on AI governance.

“By connecting these different efforts, we can strengthen their collective impact and give our cooperation a wider reach,” he said.