Deepfakes are unstoppable: local models and mass generation outpace any law, warns AI insider
LinusEkenstam · x · 2026-09-08
Linus Ekenstam posted a long thread warning that deeply disturbing deepfake content is now trivially easy to generate and there is no clear pathway to stopping it.
Key points:
- Laws and filters won't work: tools can be downloaded and run locally, outside any cloud guardrails, or run on rented GPUs for mass distribution at superhuman speeds.
- No forbidden words to filter: whatever can be imagined will be made, with or without consent.
- Likeness licensing might help but won't stop the floodgates; local bad actors remain uncontrollable.
- He argues future "holodeck" media will be the largest entertainment revolution ever, for good and bad, already surpassing human hand-made content in scale and naturalness.
He refuses to share how it was done to slow adoption and buy time for the conversation, asking: can we even legislate this?
More from Safety
- ShinyHunters threatens Florida DMV breach but proof is expired Epstein record — TechNadu · 2026-09-08
- Export Controls Working? H200 Sells for 280 and B300 for 450 Overseas — teortaxesTex · 2026-09-08
- Reddit asks: how would a China-US AI safety pause even work in practice? — sunstersun · 2026-09-08
- LG Smart TVs found uploading ~4GB of ACR data monthly and scanning every device on your network — ssh4net · 2026-09-08
- Beyond 'approve this tool call?': dev explores policy enforcement and intent checks for coding agents — Independent_Bag_2904 · 2026-09-08
- Humanos lets agents prove authorization before moving money with signed mandates — dscape · 2026-09-08