Photo attributed to a Spanish politician: AI, lookalike or deepfake

Photo attributed to a political leader: AI, lookalike or deepfake

An image attributed to a political leader is circulating with no one having provided a shred of evidence about its origin. Three explanations compete in the thread and none has won out: that it is a synthetic generation, that it is an authentic photograph of another woman with a similar look, or that it is a montage using the leader's face. No metadata, no original file, no link to the first source. In that vacuum, everyone supplies their own certainty.

How do you tell a real photo from an AI-generated one?

The first argument that comes up is imperfection. Some rule out artificial generation by clinging to the unflattering details of the body in the picture: asymmetries, textures, flaws that, according to this theory, a trained model tends to smooth over. Imperfection would thus be the signature of the real. Imperfection as proof of authenticity, which is rather amusing.

The reasoning holds up poorly on both sides. A current generator can produce very irregular images, and a phone camera in full sun doesn't work miracles either. The only thing that could be verified is what's missing: no one has produced the file exactly as it came out of the device, with its embedded data. One participant settles the matter with the caption "100% real no fake" and a link to MEGA, the classic formula for a file that never appears.

The ear as a clue

Anatomical comparison has taken over. One participant compares the image with other photographs of the leader and claims the ear does not match the one she is known to have. It's the kind of detail no trained eye looks at in an ordinary photo, and yet it decides these matters. The most cautious conclusion, suggested by another participant, is the reasonable likeness: a different woman who looks similar enough for the attribution to stick.

No one here has carried out that check with originals or verified files; the attribution remains unconfirmed.

The deepfake: more work and no evidence

The third route is deliberate manipulation. A deepfake requires sufficient source material, computing time and a result that holds up under zoom; it's not the same as typing a sentence and waiting for four variations. Being technically possible doesn't make it what actually peine.

On this point not a single clue has been offered. No capture of the alleged editing, no intermediate crop, no identifiable source account. Only the intuition that the result "gives it away" or that it "doesn't give it away".

From authenticity to climbing the party ladder

Halfway through the conversation, the image fades into the background and another issue appears: how one advances within a political party. Some claim, without offering any data, that the leader must be "on some party committee trying to climb the ranks" and that you have to work hard in politics to get ahead. It's speculation, of the same kind as the pixel talk, and it works just as well.

That's where the analysis stalls. With three incompatible hypotheses, zero material evidence and a discussion that is no longer about the image but about a leader's career, no one can close the matter. All that's missing is for someone to publish the original file.