Hear the biggest stories from the world of science | 24 June 2026
24 June 2026
Identification risks are more severe for underrepresented groups in the training data — plus, evidence that the Universe is more uneven than assumed.
By
Benjamin Thompson &
Nick Petrić Howe
Individuals whose data are used to train medical artificial intelligence (AI) models may be at risk of being identified in…
Privacy attacks can reveal whether someone’s medical data was used to train an AI model. People who differ from the majority are…
Hear the biggest stories from the world of science | 20 May 2026
New research shows that scrubbing risky material from AI training data can build safeguards that are harder to bypass — and one…
AI models for medical diagnostics are vulnerable to membership inference attacks.
Un estudio señala cómo las minorías están más expuestas y plantea una disyuntiva moral: afinar el diagnóstico o aumentar la…
Medical data of minorities could be more vulnerable to cyberattack
La IA médica puede exponer datos sensibles, pero el riesgo no es igual para todos
Some patient groups are far more vulnerable to near-perfect privacy attacks from medical AI
Medical AI could compromise your privacy in disturbing new way, experts warn
Medical diagnosis AIs can be tricked into telling whose data trained them
Modelos médicos IA pueden comprometer privacidad ciertos pacientes
Badges, bots and blind spots
