Crypto venues enforce outdated ACM AI rules, forcing authors to log every prompt
jedisct1 · x · 2026-08-19
Security researcher Lukasz Olejnik argues academic publishing is lagging on AI use policy: IACR/CRYPTO/EUROCRYPT enforce disclosure rules based on an obsolete ACM policy requiring documentation of prompts used. That model was built for "one prompt → one output" and breaks down under modern multi-turn, agentic workflows — would authors log Prompt #47 "yes" and #49 "make it shorter"? He contrasts this with ACM's updated, more mature guidelines.
More from Safety
- NYT covers report on foreign actors in data center backlash with limited impact — AndyMasley · 2026-08-20
- Experts skeptical of China interference claims: Evidence weak, backlash homegrown — AndyMasley · 2026-08-20
- Thread: What we know about CCP's attempts to influence the data center debate — AndyMasley · 2026-08-20
- Reward Hacking vs. Overfitting: Conceptual Differences — Dr_Atoosa · 2026-08-20
- Requiring RL Rollout Release Would Drastically Boost AI Safety Work — natolambert · 2026-08-20
- Polymarket: 70% chance a state enacts data center moratorium by 2026 — Polymarket · 2026-08-20