Evidence file

How independent is Wenzu The Voice?

In the election campaign he posted 92 reels favouring Labour and 0 criticising the government alone. He made 42 promo reels for government schemes, with 0 ad labels.

Last updated on October 12th 2026 · 1,153 reels, 802 transcribed · Compiled by Miirko Sant

Wenzu The Voice tells his viewers he is on no side: 65 times in the reels studied here he calls himself neutral or independent, or says he favours neither party. This file tests that promise against what he actually posted. It reads 1,153 of his Facebook reels, from May 2024 to October 2026, classifies each one, and codes every praise and criticism of Labour and the Nationalist Party (PN) the same way.

The answer is the same on every test. He criticises the PN almost three times as often as Labour, and differently. In the five-week election campaign he did not post one reel criticising the government on its own. He promoted government schemes in reels that carried no ad label, and almost all the insults he aimed at politicians went to one side. The charts and quotes below show the evidence; each quote links to its reel. ■

65times he says he is neutral, independent or on no side 245:91criticisms of the PN and of Labour 498,194views on his 42 unlabelled promo reels 60:5insults aimed at PN and at Labour politicians

Download the booklet (PDF)Explore all 406 political reels Get the data

The case in one page

A partisan, not an independent

The argument

Every test gives the same answer. He goes quiet on Labour when the vote is near, treats the two parties by different standards, performs at Labour's congress, sells the government's schemes without a label, insults one side, and makes rules only for his opponents. Liking Labour is his right. Calling that independence is not honest.

Part	The point	Key number
1. What he says about himself	His whole brand is 'I am on no side'. Every other part of this file tests that promise against what he actually did.	65 times he says he is on no side
2. The drift	When the vote came near, he stopped criticising Labour on its own: 15 such reels in his first two months, zero in the campaign. A neutral voice gets sharper at election time; a partisan goes quiet.	Campaign: 92 reels for Labour, 0 against the government alone
3. Two standards	He does not only criticise the PN more often; he criticises it differently. Names and ridicule for the PN, 'il-gvern' and a 'but' for Labour. That is a partisan's double standard.	Ridicule in 132 of 245 PN criticisms (54%), 15 of 91 Labour (16%)
4. Campaign production	He made campaign content at Labour's congress with Labour ministers, and 30 of the 31 reels with a politician on camera are with Labour. A commentator puts both sides on camera; he performed for one.	30 of 31 reels with a politician on camera are with Labour
5. Government ads	He sold the government's schemes in 42 reels, in the government's own words, attacked the PN figures who questioned them, and labelled none. That is the government's voice, not an independent one.	42 promo reels, 498,194 views, 0 labels
6. “I never insult”	His civility rule runs one way: 60 insults at PN politicians, 5 at Labour ones. The rule he preaches protects Labour, not everybody.	60 insults at PN politicians, 5 at Labour
7. His own rules	He tells others not to tell people how to vote, then steers his own viewers away from the PN two days before the vote. His rules bind his opponents, not himself.	Same day, 28 May 2026

Not argued: that he likes or votes Labour (his right), or that he was paid (no evidence). The case is that he presents himself as independent while acting as a partisan.

Discover more

Method

A note on method

1. His promise

What he says about himself

2. The drift

The drift: his criticism disappears when it counts

3. Two standards

Two standards: how he treats each party

4. On camera

Campaign production

5. Government ads

Government ads and disclosure

6. Insults

“I never insult anyone”

Methodology

The audio of each reel was transcribed by an AI speech-to-text model (ElevenLabs Scribe) and classified by an AI model (Claude, by Anthropic) into eight categories, then re-reviewed in full several times. Maltese transcription makes mistakes, so each quote links to its reel. Reels over ten minutes were not transcribed. Read the full method.