AI may need aviation-style safety rules to earn trust, Josephine Teo says

Josephine Teo, a minister in Singapore's government, said artificial intelligence may need aviation-style safety regulation before the public can trust it. The Straits Times reported the remark, making her one of the latest officials to compare AI governance with a mature, safety-critical industry.

Aviation is a recurring analogy because it turned a high-risk business into a manageable one: aircraft need airworthiness certification, pilots need licences and logged hours, accidents must be reported and traced to a cause, and the rules are aligned across countries through international bodies. That culture of certification, record-keeping and accountability is what the comparison points to.

Translating it to AI is not straightforward. A model's behaviour cannot be stress-tested the way an airframe can, capabilities drift with each release, and deployment contexts are highly fragmented, so the same model in healthcare, finance or industrial control presents an entirely different risk profile. Aviation-style oversight is therefore more likely to arrive as tiered, sector-specific compliance duties than as a single licence.

The weight of the remark comes from where it was made. Singapore has long attracted AI investment with light-touch rules and regulatory sandboxes, while still guarding the line in sensitive areas such as finance and healthcare. A minister speaking publicly about tougher safety rules suggests the trust deficit has become a constraint the industry has to address to keep expanding.

The comment also fits the wider regulatory rhythm. The European Union assigns obligations by risk tier, the United States leans more on existing agencies and self-regulation, and several Asian economies favour building testing and evaluation capacity alongside adoption. The value of the aviation analogy is that it supplies a shared reference point: trust depends on being able to locate a cause and assign responsibility when something fails.

What to watch is whether Singapore turns the remark into institutions, such as assessment and registration requirements for high-risk uses, incident-reporting mechanisms, and work on international mutual recognition. As an analogy alone it constrains little; once it enters legislation and enforcement, compliance costs and release timelines across the region get recalculated.