ChatGPT Prompts for LinkedIn Headshots: 20 Professional Edits That Still Look Like You

The short version: you don't need a £300 photographer or a fake-face AI headshot app to fix your LinkedIn photo. Upload a real photo of yourself into ChatGPT and use targeted edit prompts, one for the background, one for the lighting, one for the collar that's gone wonky, and you'll get a headshot that looks like a better version of you rather than a stranger wearing your name.

Useful alongside this: How to Write Aesthetic Photo Prompts for AI Image Tools.

Why editing your own photo beats generating a new one

I'll say the bit most posts on this topic skip over. Those AI headshot generators that ask you to upload twenty selfies and spit out a "professional" version? They're often making a face that is close to yours but not yours. I've had three people this year tell me a client or a former colleague did a double take when they met them in person because the LinkedIn photo looked like a slightly different person, better jawline, smaller nose, eyes a shade bluer. That's not a flattering headshot. That's a trust problem before you've even said hello.

ChatGPT's image editing is different, and it's the reason I now use it instead of the dedicated headshot apps. You give it one real photo and it adjusts what's around your face, background, lighting, wardrobe colour, posture, without redrawing your features. The result still passes the "does this look like the person who walked into the meeting" test, which is the only test that matters on a networking platform.

My own headshot, and what changed

My LinkedIn photo until last spring was from a 2019 conference in Tel Aviv, taken by a colleague on her phone under the venue's orange-tinted downlights, with a fire exit sign glowing behind my left shoulder. I knew it was tired. I hadn't got round to fixing it because a proper photoshoot felt like a whole afternoon I didn't have.

I uploaded that same photo into ChatGPT and ran it through four prompts in one sitting: a background swap to plain charcoal grey, a white balance correction to pull out the orange cast, a shine reduction on my forehead (conference lighting is not kind to a 53 year old forehead), and a light sharpen on the jacket lapel that had gone soft and blurry from the phone camera. Twelve minutes, no cost beyond my existing ChatGPT Plus subscription, and the fire exit sign was gone. Compare that to the £250 branding photoshoot I paid for in 2022, half a day out of my diary and a photographer who kept telling me to "relax my shoulders" for two hours.

The point isn't that AI editing replaces a real photographer forever. It's that for the 90% of people whose LinkedIn photo problem is bad lighting, a messy background, or a photo that's five years and one haircut out of date, this fixes it in less time than it takes to make a coffee.

How to run these prompts

1. Open ChatGPT (the free tier can do basic edits, GPT-4o and Plus give you more consistent results on skin tone and background blending).
2. Upload a recent photo where your whole face is visible, taken in good light if possible, even a phone photo by a window works.
3. Paste one prompt at a time from the list below rather than stacking five changes into one request. ChatGPT handles single, specific instructions far better than a paragraph of demands.
4. Look closely at hands, ears, and collar edges after every edit. These are where AI image tools slip.
5. Ask for one more pass if something's off: "keep everything the same but reduce the shine on my nose slightly."
6. Download at the highest resolution offered, then crop to LinkedIn's recommended 400 x 400 pixel square before uploading, since LinkedIn displays it at 200 x 200 on desktop and will compress a bad crop badly.

If you want the wider mechanics of how to structure prompts so ChatGPT follows them instead of guessing, I've broken that down in this guide to ChatGPT prompt examples.

Background and setting edits

• 1. Studio grey swap: "Replace the background with a smooth, neutral charcoal grey studio backdrop, softly out of focus, keep my face, hair and clothing exactly as they are."
• 2. Office declutter: "Remove the bookshelf and clutter behind me and replace it with a softly blurred, plain office background in a warm beige tone."
• 3. Outdoor to indoor: "Change the outdoor background to a softly blurred indoor setting with warm natural light, similar to a modern office lobby."
• 4. Brand colour backdrop: "Replace the background with a solid backdrop in a deep navy blue that complements my jacket, keep the edges of my hair natural and unedited."
• 5. Depth of field fix: "Blur the background slightly more, as if shot with a wider aperture, so my face is the clear focal point."

Lighting and colour edits

• 6. White balance correction: "Correct the colour temperature so the lighting looks neutral daylight rather than orange or yellow toned."
• 7. Shine reduction: "Reduce the shine and glare on my forehead and nose without changing my skin texture or smoothing out my features."
• 8. Shadow lift: "Lift the shadows slightly under my eyes and jawline so the lighting looks more even, keep it natural, not airbrushed."
• 9. Soft key light effect: "Add a soft, even front-facing light as if from a large studio softbox, reduce harsh side shadows on my face."

Wardrobe and styling edits

• 10. Colour correct the jacket: "Change the colour of my blazer from black to a deep burgundy, keep the fit and fabric texture exactly the same."
• 11. Fix a collar: "Straighten my shirt collar so it sits evenly on both sides, don't change anything else about the photo."
• 12. Remove a distraction: "Remove the lanyard visible around my neck, fill in the shirt collar naturally underneath."
• 13. Add subtle polish: "Smooth a few loose strands of flyaway hair without changing my hairstyle or hair colour."

Retouching and skin edits

• 14. Natural skin evenness: "Even out my skin tone slightly and reduce redness around my nose and cheeks, keep visible pores and natural texture, this should still look like a real, unfiltered photo."
• 15. Under-eye fix: "Reduce the darkness under my eyes slightly, keep it subtle and realistic, don't smooth my skin elsewhere."
• 16. Remove a temporary blemish: "Remove the small spot on my chin, blend it naturally with the surrounding skin tone."

Expression and pose edits

• 17. Warmer expression: "Adjust my expression very slightly so I look a touch more relaxed and warm, keep my face shape and features exactly the same, this should still clearly look like me."
• 18. Straighten posture: "Straighten my shoulders slightly so I'm facing the camera more directly, keep everything else in the photo unchanged."

Final polish and crop edits

• 19. Framing fix: "Recrop the image to a head and shoulders portrait, centred, with even space on both sides, suitable for a square LinkedIn profile photo."
• 20. Consistency pass: "Match the lighting and colour tone of this photo to a warm, professional studio look, similar to a corporate headshot, without over-processing the skin."

Run through these in order and you'll notice something: none of them ask ChatGPT to change your nose, jaw, eye colour, or age. That's deliberate. The moment a prompt starts altering bone structure, you've crossed from photo editing into face generation, and that's exactly where headshots stop passing the "does this look like the person I met" test. If you want more of this kind of restrained, identity-preserving prompt written out in full, I've put together thirty more portrait and headshot edits here, and a broader set covering everyday photo fixes in sixty copy-paste photo editing prompts.

The mistake almost everyone makes with these prompts

People stack too much into one request. "Make me look professional, fix the lighting, change the background, and make my smile better" produces a muddled edit where ChatGPT guesses at priorities and usually overcooks the skin smoothing because that's the easiest instruction to satisfy. Run one change, check it, run the next. It takes longer to type six separate prompts than one long one, but you end up with a photo you'd be comfortable defending if someone asked "is that real?" And on LinkedIn, where your photo sits next to your name for years, that matters more than shaving four minutes off the process.

Worth knowing too, if you're chasing a specific aesthetic rather than just fixing flaws: there's a wider trend right now of people using ChatGPT for stylised portrait looks that go well beyond a standard headshot. I've covered what's popular and how to avoid the generic AI face problem in this guide to striking portrait prompts, worth a read if a plain corporate headshot isn't the vibe you're going for.

If you want your own numbers, use the free prompt generator.

I keep every related walkthrough in the ChatGPT Prompts and Limits: 30 Guides to Fun Prompts, Photo Edits and Daily Caps.

Frequently asked questions

Is it against LinkedIn's rules to edit my headshot with AI?

No, editing your own real photo, adjusting lighting, background, or colour, is not against LinkedIn's terms. The line you shouldn't cross is uploading a fully AI-generated face that isn't recognisably you, since that misrepresents who someone is connecting with.

What's the best photo to start with before using these prompts?

A recent photo, taken within the last year, with your whole face visible and reasonable lighting, even a phone photo by a window. ChatGPT can fix lighting and background well but it can't fix a photo where your face is mostly in shadow or turned away from the camera.

Do these prompts work in the free version of ChatGPT?

Most of the simpler edits, background swaps, colour correction, collar fixes, work fine on the free tier. Skin and lighting edits that need to preserve fine detail tend to come out more consistent on GPT-4o with a Plus subscription.

How do I make sure the edited photo doesn't look overly airbrushed?

Add "keep visible pores and natural skin texture, this should still look like a real, unfiltered photo" to any skin-related prompt. Without that instruction, ChatGPT tends to default to smoothing skin more than a genuine professional headshot ever would.

Further reading

• LinkedIn's Help Centre
• OpenAI's own platform documentation