On Anwar’s AI, Cambridge Analytica, and the data we stopped caring about giving away.
What Most People Saw
When the Anwar AI launched, most Malaysians dismissed it as the government joining the AI bandwagon — a novelty chatbot dressed in political branding, irrelevant to anything that actually matters.
That reaction missed what was actually happening.
The Power Distance Illusion
Ask yourself an honest question. When was the last time you had a burning question for your leader — something they genuinely needed to hear — but couldn’t ask because of the massive power distance between a citizen and a prime minister?
That distance is real. It shapes what gets said publicly. What gets suppressed. What frustrations never make it out of WhatsApp groups and into actual policy conversations.
PMX AI appeared to solve this. The power distance felt gone. You were typing directly to your PM. No gatekeeper. No protocol. No fear of consequence.
But the wall didn’t disappear. It got friendlier. That’s not the same thing.
You’re not talking to Anwar. You’re talking to a model trained to sound like Anwar. Your frustration is acknowledged. Your question is answered — or deflected. And then it disappears into a database.
No direct line to policy. No guarantee anyone reads what you wrote. No feedback loop that connects citizen sentiment to government action.
It’s a venting machine. Not a listening mechanism.
The Cambridge Analytica Parallel
If you’ve watched The Great Hack — the Netflix documentary on Cambridge Analytica — you already know Malaysia isn’t unfamiliar with this story.
During Najib’s time, the Malaysian government was connected to Cambridge Analytica’s operations. Data harvested from Facebook profiles without explicit consent. Micro-targeted political messaging. Voter behaviour manipulated at scale.
Cambridge Analytica needed a third-party app, Facebook’s API, 270,000 people to consent unknowingly, their data scraped to reach 87 million profiles, and a legal team to argue it wasn’t illegal.
PMX AI needed none of that. It needed one chatbot. A government brand that feels trustworthy. Citizens to come voluntarily. Terms and conditions nobody reads. And zero legal exposure — because the fine print almost certainly contains a clause that says conversation data can be used for service improvement, research, or unspecified purposes.
Cambridge Analytica had to steal what people didn’t knowingly give. PMX AI gets people to give it freely — while thinking they’re just asking their PM about the price of eggs.
Same data. Different legality. Infinitely more scale.
The Sentiment Harvesting Layer
Every question asked — catalogued by topic, by concern, by urgency. Every frustration expressed — mapped by region, by demographic, by language used when people are unguarded. Every repeated question from the same area — a signal that something is wrong there that the official channels haven’t surfaced yet.
This isn’t a public service tool. It’s the most sophisticated political polling operation ever built — with citizens voluntarily providing the data, in their own words, completely unprompted.
Which state is angry about what. Which demographic is worried about which issue. What language Malaysians use when they think they’re just talking to their PM. All of it — campaign intelligence.
The Door We Left Open
People have been warned. Repeatedly. For years.
The warnings came. People heard them. And then kept scrolling. Kept posting. Kept clicking agree on terms and conditions they never read.
The repeated warnings didn’t create vigilance. They created fatigue. And fatigue looks exactly like consent.
By the time PMX AI arrived — the door wasn’t broken open. It was already wide open. Because the public had been so thoroughly sedated by years of data privacy warnings that one more didn’t register.
Cambridge Analytica needed a scandal to expose what they were doing. PMX AI needed nothing. Just a launch event. A press release. And a public that had long since stopped caring about the distinction between giving data and having it taken.
They didn’t break down the door. They waited until we stopped caring that it was open.
The Question Worth Asking
Cambridge Analytica was a scandal because they took what people didn’t knowingly give.
What happens when people give it freely — without knowing that’s what they’re doing? What happens when the terms and conditions nobody read contain the clause that makes it all legal? What happens when the sentiment harvested from millions of conversations becomes the foundation of the next election campaign — and nobody can call it a scandal because technically everyone consented?
That’s not a hypothetical. That’s the architecture already in place.
A Note On Timing
This briefing was written when PMX AI — Anwar Ibrahim’s AI digital twin — launched on 18 July 2026. It was taken offline less than 24 hours later, officially for system upgrades.
The architecture described here remains relevant regardless of whether the chatbot is live or not. And it became more relevant on 10 September 2026, when Anthropic published a threat intelligence report confirming that a commercial operation had deployed over 1,000 fake AI-powered social media accounts to manipulate Malaysian political sentiment — constituency by constituency.
One side collects what citizens genuinely feel. The other manufactures what citizens appear to feel. Both are happening in the same political cycle.
That gap is worth understanding before the next election.
lokman@objektif.cc
Clarity is the most underrated business investment.
— Lokman S., Majalah BIKIN