AI Made Malaysia's Prime Minister Endorse Scams He Had Never Heard Of
What happened
Malaysia reported 13,956 investment fraud cases in a single year tied to a single method: AI-generated deepfake videos placing the faces and voices of the country's most recognizable leaders into fake endorsements. Prime Minister Anwar Ibrahim and other officials, political figures, and business leaders appeared in footage they never filmed, urging viewers to put money into schemes that did not exist.
The videos circulated through Facebook, WhatsApp, and Telegram, platforms that reach tens of millions of Malaysians daily. The endorsements looked and sounded real. Viewers saw a head of government or a familiar business figure speaking directly to camera, naming a platform, promising returns, and directing them to invest. The fakeness was not obvious, and by the time takedown requests succeeded on any given piece of content, copies had already spread beyond the originals.
Scammers proved fast enough to outlast every platform response. When deepfakes were removed, new ones replaced them. Authorities attributed RM2.11 billion in losses to these schemes over the reporting period, a figure that reflects only reported cases. The actual number of victims is almost certainly higher because investment fraud in general goes underreported, and fraud involving material that has already been deleted from the platforms is harder to trace after the fact.
The mechanism that makes deepfake investment fraud effective is not technical complexity. It is the trust that flows from a familiar, authoritative face. A viewer who sees a sitting prime minister describe an investment opportunity does not start from skepticism. The social credential of a recognizable leader is borrowed without consent and deployed in a context where it does maximum persuasive damage: low-cost video distribution with no verification layer at the point of distribution.
What scams at this scale expose is an absence of reliable provenance for synthetic media. A viewer has no way to verify, at the moment a video arrives on their phone, whether the face in it ever recorded it. Platforms can remove content once it is flagged, but removal after the fact does nothing for the person who already transferred money. A provable record of what a system produced, tied to whoever authorized its release, would not stop every scam, but it would shorten the window in which fabricated endorsements travel unchallenged. Right now that window is wide enough to cost more than two billion ringgit in a single year.
Reported impact
- Affected parties
- Not publicly disclosed
- Harm type
- Not publicly disclosed
- Scale
- Not publicly disclosed
- Financial impact
- Not publicly disclosed
- Regulatory action
- Not publicly disclosed
Classification
Relevant governance controls
Governance control mapping is not available for this record.
- No controls mapped
Not publicly disclosed
Control mapping is analytical. It does not state that any control would have prevented the incident.
Sources and evidence
This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.