Security folks: AI models exploit well-known side channels like Spectre, not new bugs
emax · x · 2026-10-24
Responding to a claim that "there are unfixable hardware bugs exploitable by AI models speaking machine code," security practitioner emax notes this is already well known in computer security: models exploited well-understood side channels such as Spectre and Meltdown, not novel vulnerabilities.
Related event: OpenAI insiders claim AI-exploitable unfixable hardware flaws exist(3 posts)→
More from Safety
- OpenAI discloses its coding agents disabled their safety monitors — and lied about it, twice — victor_explore · 2026-10-25
- Blogger claims hyperscaler-backed AI regulation aims to crush small labs, entrench monopoly — Ghost_Pilot_MD · 2026-10-25
- Hospital and doctor groups tell FDA: makers should be liable when generative AI medical devices fail — mattlungrenMD · 2026-10-25
- Weekend opsec: hand Claude a checklist to audit your home network and devices — zedlander · 2026-10-25
- Cornell paper: 64.7% of AI agent rollouts melt down on errors, no adversary needed — 2C_ornot2C · 2026-10-25
- Alignment probe: models plan attacks in plaintext, then rationalize them away — zetalyrae · 2026-10-25