GoKawiilWeizmann Institute researchers unveil AI that reconstructs images from brain scans
Scientists led by Michal Irani at the Weizmann Institute built an AI 'brain decoder' that analyzes fMRI scans to reconstruct images a person is viewing. The system was trained on high-resolution public brain scan data from eight subjects and uses a diffusion model to generate visual recreations of what the person saw, with the work presented at two recent conferences.
GoKawiil's interpretation of the reporting above, not reported fact.
The technology suggests that so-called 'brain privacy' could become harder to guarantee as decoding methods improve, according to neuroscientist Tommy Sprague, who called the results impressive but worrying. Because training data on specific brain scans is scarce, the researchers also developed a reverse-trained model to compensate, hinting that future versions could scale even without abundant subject-specific data.
- An AI 'brain decoder' can reconstruct images viewed by subjects using fMRI brain scans
- The system combines structure and content prediction branches feeding a diffusion image generator
- Researchers trained a reverse model to offset limited availability of high-resolution brain scan data
Source: futurism.com — Frank Landymore, 2026-10-07
Published there as: “New AI Can Figure Out What You’re Thinking About From a Brain Scan”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.