Crime's new toolkit

The room

The Asia Region Law Enforcement Management Program (ARLEMP) has trained over 1,200 law enforcement officials from 44 countries since 2005, delivered in partnership between the Australian Federal Police, Vietnam's Ministry of Public Security, and RMIT University Vietnam.

In June 2026 I was invited to deliver a one-hour AI capability and threat session to around thirty senior participants from across ASEAN — commissioners, regional commanders, and senior investigators from Thailand, Laos, Cambodia, the Maldives, the Philippines, Indonesia, Vietnam, and Australia, alongside a representative from the United Nations Office on Drugs and Crime.

This was not a higher education audience.

The challenge

The design problem was transfer: taking a framework built in and around higher education and making it land immediately for an audience whose professional world — evidence, prosecution, organised crime, border security — has almost no surface overlap with university governance.

The session started late, with an audience of unknown AI familiarity. The content had to be flexible, hands-on, and immediately recognisable as relevant to investigators and commanders.

Making the gap visible

Rather than asserting that AI had advanced rapidly, the session opened by surfacing the room's own assumptions. Participants were asked in sequence: do you interact with AI by typing? By phone? Did you know AI can speak? Did you know it can see?

On the third question, roughly half the room indicated they were unaware AI could speak in real time. Fewer still knew it could see. The capability–perception gap was demonstrated live, before a single slide had been discussed. Live polling confirmed it: two-thirds of respondents placed their institutions at "Experimenting" — the earliest stage of AI adoption. Only two placed themselves at "Embedded."

The demonstration

A live session with Google AI Studio showed the room an AI that could see, speak, and translate in real time. The convener confirmed that te reo Māori is rarely encountered outside New Zealand — the AI read a Māori document aloud, fluently. A Mandarin speaker addressed it in Chinese; it responded and translated. A Cambodian police officer stood up and spoke to it in Khmer. It answered correctly.

The room shifted from reception to participation.

Deepfakes: the evidence crisis

Participants first played a detection game — attempting to distinguish AI-generated images from real ones. Even attentive, trained observers struggle to identify synthetic media reliably. They were then shown how easily convincing synthetic media can be created. No technical skills. No special equipment. Seconds.

The operative concept was the liar's dividend: even if a deepfake is never used offensively, its existence lets guilty parties cast doubt on authentic evidence. The defence doesn't need to prove the evidence is fake — only raise the question. At least two senior participants stated explicitly that their agencies had no systems for detecting or managing AI-generated evidence.

The agentic threat

The final section moved from current capability to near-term adversarial risk — the distinction between individual criminal AI use and coordinated agentic use: autonomous systems, networked agents, simultaneous multi-target campaigns operating at machine speed, across jurisdictions, without fatigue.

The framing: this is not an IT problem. It is a leadership problem. That framing was validated publicly in the room by the UN representative.

Three questions before any agency adopts any tool:

• What data does it touch?
• What content does it see?
• What can it do when nobody's watching?

These are not IT questions. They are accountability questions.

What people said

"Sir Nick is very, very good. He is one of a kind, a futurist. You cannot find someone like him in the Philippines."

"Everyone's still talking about it."

What this demonstrates

Commissioners and regional commanders from across ASEAN are not a forgiving audience for abstraction. They left with a specific conceptual vocabulary, a live demonstration of AI capability most had not previously encountered, and an explicit acknowledgment of the gap between where AI-enabled criminal operations are heading and where most agencies currently sit.

The framework travels. The evidence is the room.