TL;DR: Key Takeaways
- Most online skin cancer images show one thing: a dark mole on pale skin. That excludes the majority of the world's population.
- On brown, dark brown and black skin, melanoma often appears on palms, soles and under nails — places almost no one checks.
- Late diagnosis is far more common on darker skin, leading to worse survival outcomes.
- Many AI screening tools share the same bias as the textbooks. The fix is training and validating across all Fitzpatrick types.
- ScanSkinAI is built and validated across Fitzpatrick I–VI with 95%+ accuracy as a baseline for every skin tone.
The Problem No One Talks About
Search "what does skin cancer look like" and you'll see dozens of images. Almost all of them show the same thing: a dark, irregular mole on pale, white skin.
If your skin is brown, dark brown, or black, those images are almost useless to you. Skin cancer doesn't look the same on every skin tone — and the fact that most health resources pretend it does is one of the biggest gaps in skin health education today.
This isn't just a representation problem. It's a medical one. When people with darker skin tones can't recognise the warning signs because every reference image looks nothing like their skin, they're more likely to miss early symptoms. And when skin cancer is caught late, outcomes are significantly worse. Our deep dive on skin cancer in skin of colour covers the survival-rate gap in detail.
The Skin Tone Gap in Skin Cancer Awareness
Here's what the statistics tell us:
People with darker skin are less likely to be diagnosed with melanoma overall — but when they are diagnosed, it's far more likely to be at an advanced stage. Late-stage diagnosis means lower survival rates, more aggressive treatment, and worse outcomes across the board. The 2026 global skin cancer statistics show this disparity clearly.
Why? Because nearly everything about skin cancer awareness — from public health campaigns to the ABCDE self-check method to the image databases used in medical education — was developed using lighter-skinned populations. The warning signs people are taught to look for (dark moles, colour variation against pale skin, visible border irregularity) simply present differently on melanin-rich skin.
On darker skin tones, skin cancer often appears:
- In different locations. Melanoma in people with darker skin is more likely to appear on the palms of the hands, soles of the feet, and under fingernails or toenails — places most people never think to check. This subtype, called acral lentiginous melanoma, is the most common form of melanoma in Black, Asian and Hispanic populations. See our guide to melanoma on nails.
- With different visual cues. Instead of the classic 'dark mole on light skin,' skin cancer on darker skin may appear as a dark streak under a nail, a non-healing sore on the foot, or a patch that looks slightly different from surrounding skin — subtle changes that are easy to dismiss.
- Without the 'classic' warning signs. The ABCDE checklist (Asymmetry, Border, Colour, Diameter, Evolving) was designed around how melanoma presents on lighter skin. Many of these criteria are harder to assess on darker skin, where natural pigment variation is normal and colour contrast is less obvious.
For more on these subtypes, read our guides on melanoma on nails (acral melanoma) and amelanotic / pink melanoma, both of which are commonly missed across all skin tones — but especially in patients of colour.
Get a free AI skin check — built for every skin tone
Photograph any mole, spot, nail change or non-healing sore. ScanSkinAI is validated across Fitzpatrick I–VI and screens for 80+ conditions.
Why Most Skin Screening Apps Have the Same Problem
If the medical textbooks are biased, it shouldn't be surprising that many AI skin screening tools are too. Most AI models in dermatology were trained predominantly on images of lighter skin — because that's what the available medical datasets contain. Our explainer on how accurate AI skin scanning really is goes deeper into where these models succeed and fail.
The result is a technology gap that mirrors the education gap: AI tools that work well for fair-skinned users but perform inconsistently — or poorly — for people with darker skin tones. If the training data doesn't represent you, the technology won't work reliably for you.
This matters because AI skin screening is increasingly positioned as a first line of defence — an easy, accessible way to check your skin without needing an immediate dermatologist appointment. That promise only holds if the AI actually works for everyone, not just the populations that are already best served by the existing healthcare system.
What Equitable AI Skin Screening Looks Like
Fixing this problem isn't just about adding a few more images of darker skin to a dataset. It requires a fundamentally different approach to how the AI is built, trained, and validated. Curious how it works under the hood? See how AI skin analysis works.
At ScanSkinAI, equitable accuracy isn't a feature we added after the fact. It's a design principle that shapes every layer of the technology:
Training data that represents all skin tones
Our datasets are curated to include clinical images across all six Fitzpatrick skin types — from very fair (Type I) to very dark (Type VI). This isn't a token effort; it's a deliberate strategy to ensure the model learns what skin conditions look like on real, diverse skin. Not sure of your own type? Take our free Fitzpatrick skin type quiz.
A two-tier AI architecture built for precision
ScanSkinAI uses a DINOv2-based visual classifier as its first layer, followed by LLM-powered clinical arbitration as a second layer. This two-tier approach means the system doesn't rely on a single model's judgement — it cross-references findings to reduce errors, particularly in cases where skin tone could affect visual assessment. Compare this with what AI catches that the human eye misses.
Clinical validation across all Fitzpatrick types
Our clinical validation programme doesn't just report an overall accuracy number. It breaks results down by skin type, ensuring that performance is consistent whether your skin is Fitzpatrick I or Fitzpatrick VI. The result: 95%+ accuracy across all skin tones — not as an average, but as a baseline for each group.
80+ conditions, not just melanoma
Skin health isn't only about cancer. People with darker skin tones are also affected by conditions like hyperpigmentation, keloids, dermatosis papulosa nigra, and many others that are underrepresented in standard dermatology tools. ScanSkinAI screens for over 80 conditions, giving you a comprehensive skin health picture. See the full list of skin conditions AI can detect, plus our guides on dark spots and hyperpigmentation and skincare for dark skin tones.
What You Should Actually Check For
If you have a darker skin tone, here's what to add to your skin self-check routine beyond the standard ABCDE method. Pair this with our step-by-step skin self-exam for the full monthly routine.
- Check your palms and soles. Run your fingers along the soles of your feet and look at your palms carefully. Any new dark spots, non-healing sores or areas of thickened skin deserve attention.
- Look under your nails. A dark streak running lengthwise under a fingernail or toenail — especially one that's new, widening or on a single nail — should be checked by a doctor. This can be a sign of subungual melanoma.
- Watch for non-healing wounds. A sore that doesn't heal within a few weeks, particularly on the lower legs or feet, can be a sign of skin cancer — not just an injury that's slow to recover.
- Don't ignore lighter patches. On darker skin, some skin cancers may appear as lighter or pinkish patches rather than darker ones. Any persistent change in skin colour or texture is worth investigating.
- Use AI screening as your backup. Your eyes are a good starting point, but they have limits — especially when the visual references you've been given don't match your skin. An AI tool validated across all skin tones acts as a second pair of eyes that's been trained on skin that actually looks like yours.
Want a structured monthly habit? Try our monthly skin check routine and consider keeping progress photos to track skin changes over time.
Skin Health Is for Everyone
Skin cancer doesn't discriminate by geography or ethnicity. From Singapore to Lagos, from São Paulo to London, people of all skin tones deserve screening tools that work for them — not just tools that were designed for one population and assumed to be "good enough" for everyone else.
The gap in skin health equity isn't going to close on its own. It closes when technology is built intentionally for diversity, when clinical validation includes everyone, and when people have access to tools that actually see them.
Your skin tone should never be the reason something gets missed.
Check your skin with AI built for all skin tones
Free skin health check, screening for 80+ conditions with 95%+ accuracy across every Fitzpatrick skin type.
Frequently Asked Questions
Frequently Asked Questions
Read next
Skin Cancer in Skin of Color
Melanoma on Nails (Acral Melanoma)
Amelanotic / Pink Melanoma
Beyond the ABCDE Method
Fitzpatrick Scale Explained
Skincare for Dark Skin Tones
Skin Self-Exam: Step-by-Step
How Accurate Is AI Skin Scanning?
ScanSkinAI is developed by Ivy AI Solutions Limited. ScanSkinAI is a wellbeing and risk awareness tool that flags risk levels and describes what findings may resemble. It does not provide medical diagnosis. Always consult a qualified healthcare professional for clinical evaluation.