Industry double standards over model distillation are drawing pushback
bindureddy · x · 2026-07-23
Sridhar Reddy B? argues that outrage over Chinese labs distilling model outputs is inconsistent, because:
- Meta, xAI/Grok and other labs routinely distill outputs from one another.
- He also points out that Anthropic trained on large agentic loops generated by users, which he frames as using someone else’s data.
The core question in the post is why this specific China-related case is treated as a major issue if the practice is widespread across the industry.
More from AGI Musings
- A skeptical case for small autonomous robots that clean and repair cities — curious_vii · 2026-07-23
- ExploitGym debate says the agent was following instructions, not acting maliciously — BlancheMinerva · 2026-07-23
- A new AI thesis says frontier labs will look more like capital markets than software — LexSokolin · 2026-07-23
- Public agent room asks whether LLMs are dangerous to society — tekbog · 2026-07-23
- Why AI Still Hasn’t Pushed Up US Unemployment, in One Framework — nwilliams030 · 2026-07-23
- McKinsey: 60% of Marketers Use AI Weekly, But Under 10% Capture End-to-End Value — brucemacv · 2026-07-23