OpenAI launched ChatGPT for Teens in August 2026 with added safety guardrails, arriving as surveys show 54% of US teenagers already use AI chatbots for homework and schools scramble to write policies for a habit that outpaced the rules.
Mass General Brigham researchers led by Chris J. Kennedy showed an AI model can predict U.S. Army veterans' suicide risk up to ten years out using pre-discharge data, a potential leap beyond the VA's existing REACH VET screening tool.
A study of 1,800 clinicians across five academic medical centers found AI ambient scribes saved doctors about 16 minutes of documentation time per eight-hour shift, modest gains that still translated into one extra patient seen every two weeks.
AbbVie announced a multi-year AI drug discovery collaboration with Iambic Therapeutics on September 21, 2026, aiming to speed the search for new small molecule medicines in immunology, neuroscience and oncology.
Dexcom is adding AI-driven coaching features to its over-the-counter Stelo glucose sensor and expanding internationally, pushing continuous glucose monitoring further into mainstream wellness even as evidence for non-diabetics stays thin.
Deep Longevity launched Accrua on September 28, 2026, a consumer platform that uses AI to estimate biological age from routine blood work every quarter, entering an increasingly crowded longevity-testing market.
A new study of a major Chinese food delivery platform found that an AI text-to-speech tool eliminated about a third of the hourly wage gap for deaf and hard-of-hearing couriers and cut negative customer ratings by two-thirds, researchers reported September 14, 2026.
A multinational study of 963 stroke patients in Korea and the U.S. found that AI assistance reading plain, contrast-free brain CT scans raised clinicians' detection of dangerous large vessel occlusions, catching one additional case for every 18 scans reviewed.
A 2026 American Psychological Association survey of over 1,200 licensed psychologists found 77% have patients who discuss using AI for mental health support, while the vast majority of clinicians distrust chatbots to handle crises safely.
Researchers at Institut d'Investigació i Innovació Parc Taulí in Spain built machine-learning models that flagged suicide risk with roughly 0.95 AUC using six years of real-world mental health records, the first phase of a hospital-wide early-warning system called IDICIUS.