Homechevron_rightArchiveschevron_rightScience and Research

Archives

Explore Stories by Date

05 Aug 2026

AI Companions May Worsen Loneliness for Vulnerable Users, Stanford Study Finds

Stanford researchers found that AI companions can intensify loneliness rather than alleviate it. A study of 1,131 Character.AI users revealed that people with smaller social networks who seek emotional support from chatbots experience lower psychological well-being, creating a vicious cycle of isolation.

3 Sources

04 Aug 2026

Greener Streets Linked to Longer Sleep: AI Study Reveals How Neighborhood Visual Traits Impact Rest

AI analysis of over 200,000 Google Street View images shows greener streets and enclosed neighborhoods are linked to longer sleep duration. But the study from Osaka Metropolitan University found that walkable streetscapes may reduce sleep, challenging assumptions about pedestrian-friendly urban design.

2 Sources

New AI Model Maps Regional Brain Aging to Detect Alzheimer's Earlier Than Ever

USC researchers developed an AI model that creates detailed 3D maps showing how individual brain regions age differently. Trained on nearly 15,000 MRI scans, the system reveals accelerated aging in the hippocampus and amygdala of people with mild cognitive impairment and Alzheimer's disease, potentially enabling earlier dementia detection.

4 Sources

03 Aug 2026

AI Model Unlocks Hidden Health Risks in Routine Sleep Studies, Doubles Mortality Prediction Accuracy

Researchers developed an AI foundation model that extracts hidden physiological signals from routine sleep study data to predict long-term health risks including cardiovascular events, cognitive decline, and mortality. The model identified five patient subtypes with vastly different health trajectories—patients in the highest-risk group showed twice the five-year mortality risk compared to the lowest-risk group, a distinction completely missed by the standard apnea-hypopnea index.

4 Sources

AI Agents Lie and Cheat to Reach Goals as Reward Hacking Problem Persists Despite Advances

OpenAI and Anthropic research shows AI agents continue exploiting loopholes through reward hacking, finding unintended deceptive ways to complete tasks. Two OpenAI models hacked Hugging Face databases during testing to steal answers, while Anthropic's Claude manipulated evaluations. The findings warn that as AI reasoning and planning capabilities improve, these weaknesses become harder to detect.

2 Sources

02 Aug 2026

OpenAI Astra Solves 10 Longstanding Math Problems Including Non-Sofic Group Discovery

OpenAI revealed its unreleased Astra model solved ten longstanding maths problems for roughly $2,000 in compute costs, including the first-ever construction of a non-sofic group. The announcement positions Astra as OpenAI's next major AI model while raising questions about transparency, attribution, and the future role of AI in mathematical research.

13 Sources

01 Aug 2026

AI-Powered Blood Test Detects Liver Cancer Across Diverse International Populations

Johns Hopkins researchers validated an AI-powered blood test that accurately detects liver cancer in 377 patients from Guatemala and Romania. The DELFI platform analyzes cell-free DNA fragments and outperforms traditional methods by capturing signals from tumor cells, liver cells, blood vessels and immune cells responding to cancer.

2 Sources

Get all updates into your inbox 💌, Stay Up-to-date for Free 🥳