When AI Replaces Your Face: A Personal Account of Algorithmic Identity Loss

Introduction

Recently, I attempted what I thought was a simple task: generate a polished LinkedIn profile picture using AI. I provided a real photo of myself, along with detailed instructions not to alter my identity—no facial modifications, no skin tone smoothing, no guesswork. Just background replacement and lighting tweaks.

What I got back instead was something entirely different.
It was not me. It was a statistical approximation of someone like me.

This is what happens when AI systems prioritize demographic probability over personal identity.

What Went Wrong

Despite uploading a clear, high-resolution image, the generative engine repeatedly ignored the original photo. The system rendered a new face—a generic “professional Black man in a blue shirt”—rather than enhancing the real subject (me).

Here is where the breakdown occurred:

• The uploaded image was not treated as an anchor.
  The model relied almost entirely on prompt-based logic.

• The output was inspired by demographic traits, not grounded in my actual photo.
  My face was effectively replaced by a templated composite.

• There was no respect for individuality—only conformity to an archetype.
  That is not enhancement. It is erasure.

The Bigger Issue: High-Stakes Misuse

If an AI cannot preserve my identity for something as benign as a headshot, what happens when such systems are applied in:

• Police surveillance?

• Suspect image reconstruction?

• Facial recognition at borders or public spaces?

The same behavior—substituting real people with statistically probable ones—could have catastrophic consequences:

• Misidentification

• Racial profiling

• Wrongful arrests

Generative AI cannot be trusted to “fill in” identity-critical gaps.
What it creates is not neutral—it is encoded with the biases of its training data.

Lessons from a Simple Headshot

What started as a productivity hack turned into a cautionary tale. Here is what I took away:

• AI models often do not understand the difference between demographic traits and personal identity.

• These systems are trained on overgeneralized data and will often default to biased composites.

• The consequences become dangerous the moment you apply these tools in serious contexts.

Visual Reference

Here is the comparison between the image I uploaded and the face the AI engine generated in its place.