Today’s edition of my newsletter just went out.
Rohan Paul Twitter · Rohan Paul (@rohanpaul_ai) · 2026-07-03
Rohan Paul's newsletter covers OpenAI's proposed 5% government equity stake, Claude Fable 5's safety router rerouting coding prompts to Opus, Meta restricting engineers from Claude Code and Codex over training data contamination risk, and NVIDIA achieving 2.42x write speed on a 30B model.
Appears in
Extraction
Topics: ai-industry-newsopenai-regulationclaude-modelsmeta-ai-policynvidia-optimization
Claims
- OpenAI has proposed giving the US government a 5% stake in its $852B business to ease regulatory pressure.
- Claude Fable 5's safety router is diverting a higher proportion of prompts—including routine coding tasks—to the more cautious Opus 4.8 model.
- Meta has restricted engineers from using Claude Code and Codex because outputs from rival AI models could contaminate Meta's own training data and create contractual complications.
- NVIDIA optimized a 30B language model to write 2.42x faster while retaining 98.7% of output quality.
Key quotes
Meta has reportedly limited engineer use of Claude Code and Codex because rival model outputs could contaminate Meta's own AI training data and create contractual trouble with Anthropic and OpenAI.
Safety Router Diverts More Prompts From Fable 5 To Opus 4.8