Ars Technica · AI· Dan Goodin·· 16 d ago
Anthropic VS code extension on Cursor backdoored for nearly 12 hours
Anthropic VS code extension on Cursor backdoored for nearly 12 hours
AI summary
It's still unknown how attackers got access to Anthropic's verified Cursor account.
Selection record
Threshold 76Media and individualsFirst 72Second 72
Not admittedSum of both 144 < twice the threshold 152
- Source tier
- Media and individuals; this tier's threshold is 76
- Pre-filter
- passed：Anthropic Claude扩展被植入后门，属AI供应链安全事件
- Same event
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Ars Technica · AI · arstechnica.com