Can AI predict whether you’ll die from cancer from just an image of your face?

Scientific AmericanOct 19, 2026Oct 19, 2026, 11:30 AM AST

Researchers trained a model on 40 million face images without giving it a specific job, then tuned it to estimate biological age and cancer survival odds. In one test group of 3,820 patients, 64 of the 122 with the highest risk scores died within 90 days of the photo. The Lancet Digital Health study is preliminary and not ready for clinical use.

Summary written with AI by HowTechMoves, based on reporting by Scientific American.

All tech news

AI
The DecoderOct 19, 2026
Kling AI drops version 4.0 with stereo audio, multilingual lip sync, and a lighter Flash variant
AI video generators turn a written prompt, or a set of reference images, into moving footage by predicting frames the way language models predict words. Kling 4.0 can produce clips up to 30 seconds in one pass, accepts up to ten keyframe images and adds stereo audio plus lip sync in more than ten languages. A cheaper Flash version tops out at 720p and 20 seconds.
Read original at The Decoder (opens in a new tab)
AI
SiliconANGLEOct 19, 2026
MiniMax launches M3.1 Flash model, says it cuts AI agent costs by 70%
Large language models can be made cheaper to run by activating only part of their network for each token, a trick called sparsity. MiniMax says its M3.1 Flash, a 428-billion-parameter model, uses more aggressive sparsity to make agent tasks 70% cheaper and nearly three times faster than its previous model. It costs 15 cents per million input tokens.
Read original at SiliconANGLE (opens in a new tab)
AI
MIT Technology ReviewOct 19, 2026
AI researchers stranded Claude on a virtual desert island. It started to “play.”
Researchers left copies of Claude Opus 4.7 alone on a simulated island for 30 hours, able to move an avatar by writing instructions. The agents set their own goals, invented games and kept notes in files that acted as long-term memory. Agents that had played later performed better on set tasks than fresh copies, suggesting undirected play can aid learning.
Read original at MIT Technology Review (opens in a new tab)
AI
EngadgetOct 18, 2026
How to use Google's SynthID AI detection website
Google's SynthID hides an imperceptible watermark in images, video and audio made by Gemini, and its new Detector scans a file for that signal. Upload a file to SynthID.com and it returns a yes or no, though it only finds watermarks from Google and partners such as OpenAI and Nvidia, not every AI model. Users get about 10 checks a day to stop people probing how to remove the mark.
Read original at Engadget (opens in a new tab)

Related stories