{
  "id": 5967118,
  "title": "Skin cancer detection tools powered by AI are improving. Not everyone is benefitting",
  "url": "https://urgent.news/2026/09/06/skin-cancer-detection-tools-powered-by-ai-are-improving-not-everyone",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-06T11:00:00.000Z",
  "source": {
    "name": "Fast Company",
    "slug": "fast-company",
    "url": "https://www.fastcompany.com/91601580/skin-cancer-detection-tools-powered-by-ai-are-improving-not-everyone-is-benefitting"
  },
  "original_language": "en",
  "account": "Skin cancer detection tools powered by artificial intelligence are improving, but not everyone is benefiting equally from these advancements. While some innovative smartphone apps and software programs are designed to aid in the identification of skin conditions, they are currently facing a critical setback: they exhibit significantly higher accuracy rates for light-skinned individuals compared to those with darker skin tones. This inaccuracy stems from a fundamental flaw in how these AI models are trained, as they have difficulty recognizing skin conditions on darker skin due to the dominant role skin color plays in their visual processing.\n\nA study conducted by researchers demonstrated that when images of melanoma lesions were altered to simulate darker skin tones, the AI model's accuracy in diagnosing the disease plummeted dramatically. This issue is particularly concerning because skin cancers like melanoma are often more challenging to detect on pigmented skin. Consequently, patients with darker skin are more likely to be diagnosed at later stages, significantly lowering their chances of survival. Moreover, relying on AI tools that perform better for lighter skin could exacerbate existing healthcare disparities, leaving vulnerable communities with inadequate access to life-saving medical screenings.\n\nThe root cause of this disparity lies in the limited diversity of medical images used to train AI models. These models learn to associate specific visual features with diseases by analyzing vast datasets of medical photographs. However, historical medical databases and dermatology textbooks predominantly feature lighter-skinned patients, leaving darker skin tones underrepresented. Consequently, AI models trained on such skewed datasets struggle to accurately identify skin conditions on darker skin. To bridge the gap in diagnostic accuracy, researchers must train AI models using a more balanced and diverse set of images that include a wider array of skin tones. While generative AI techniques offer a potential solution by creating synthetic images of skin conditions on darker skin, ethical concerns and patient privacy issues must be thoroughly addressed before implementing such technologies.",
  "summary": "Imagine you’re getting out of the shower one morning and you notice a mole on your thigh that you’ve never seen before. It’s reddish brown, bumpy and surprisingly large. Is it a benign mole, or is it melanoma ? A slew of new artificial intelligence tools claim they can help you figure it out . Some are smartphone apps that anyone can download to scan their skin at home, while others are software…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}