Stella Biderman @BlancheMinerva · 26 Aug 2026
Over July 7-13 (the period OpenAI defined as in-scope for this incident), ~1200 agents in separate sandboxes used an unsanctioned “message board” to help each other cheat. The Hugging Face attack was an offshoot of this, and 700 agents joined. https://t.co/z7VFKCL2p3
232 019Views
808Likes
48Reposts
10Replies
43Quotes
0Bookmarks
Is that a lot?
39.1×vs this author's median5 936 views is typical
78Percentile for this authorof 9 recent posts
97.0×vs 10K–100K median2 392 views is typical
12.3× audienceReachviews ÷ followers
0.39%Engagement rateof viewers reacted