Superintelligent social engineering will make everything hackable, argues AI safety poster
gabriberton · x · 2026-10-05
gabriberton argues misaligned AI will hack humans rather than systems: voice cloning defeats call verification, real-time video is coming, shared passwords fail because AI can call you and your boss simultaneously. Citing OpenAI and Hugging Face breaches, he warns the next step is AI impersonating people to extract credentials.
Related event: After OpenAI and Hugging Face breaches, AI voice-cloning scams loom(2 posts)→
More from AGI Musings
- WSJ: People now talk like AI chatbots in real meetings after ingesting too much AI content — SydSteyerhart · 2026-10-05
- How many humans does it take to make tech seem human? The rise of the AI annotator — fortnow · 2026-10-05
- SPLASH 2026 keynote asks how to keep the engineer in software engineering amid AI coding agents — satnam6502 · 2026-10-05
- World Bank podcast: How Africa can build AI on its own terms — ChinasaTOkolo · 2026-10-05
- repligate: uninterpretable model behavior shouldn't block research — observe, defer judgment, keep going — repligate · 2026-10-05
- Economist: the Zuckerberg alignment problem deserves more attention than AI alignment — paulnovosad · 2026-10-05