Rohan Paul (@rohanpaul_ai)T2·68 pts
- OpenAI's single monorepo strategy blurred visibility boundaries, creating lateral privilege escalation risks
- Internal security warnings were suppressed by competitive pressure (vs. Google/Anthropic), highlighting the speed vs. safety trade-off
- Official clarification states model weights and top secrets are excluded from the repo, attempting to mitigate perceived severity