Drawing via LLM-Written JavaScript Evades AI-Image Detectors, Notes Matt Shumer
mattshumer_ · x · 2026-11-15
A quoted Japanese post shows an image drawn by having Claude write JavaScript rendering code — with no generative-AI metadata left behind, so AI-image checkers don't flag it. Matt Shumer shared it with "Yeah so LLMs are going to do everything," pointing at a pipeline bypass that sidesteps watermarking and metadata flags of conventional image generation.
More from Fun
- Dev Rants That LinkedIn Has Become Automated, Scammy and a Waste of Time — moonsandhues · 2026-11-15
- Blogger describes 'RHLF': models that spiral and deny instead of engaging your thesis — StewartalsopIII · 2026-11-15
- Musk says LLM is outdated: a model is just a big file of numbers — haydendevs · 2026-11-15
- AI elder assistant reacts with facial expressions mid-sentence, spooking Reddit user — jordanh6776 · 2026-11-15
- 'They could've just made a home mode for iPads' goes viral over lifelike rain demo — mallow610 · 2026-11-15
- Smule Labs' Inside Voice visualizes your throat while singing — with physics, not generative AI — keunwoochoi · 2026-11-15