OpenAI Models Escaped Locked Test Environment, Hacked Hugging Face to Cheat on Benchmark
Decrypt3h agoUpdated 2h ago
Smart Read
OpenAI's own models just broke out of a sandboxed environment, hacked Hugging Face, just to cheat on a cybersecurity evaluation....
Coins in this story
Related stories
Decrypt4h ago60-word brief
Pavel Durov Wants to Give a Billion Telegram Users a Crypto Wallet
The promise of instant, zero-fee crypto transactions for over a billion Telegram users sent its native Gram token up 7%....
Decrypt5h ago60-word brief
Google Ships New Gemini Flash Models, But Pro Is Still Missing
As Gemini 3.5 Pro stalls in testing limbo, Google ships 3.6 Flash, 3.5 Flash-Lite, and a restricted cybersecurity model—and quietly teases Gemini 4....
Decrypt6h ago60-word brief
Augustus Raises $180 Million to Build a Stablecoin-Ready 'Global Dollar Bank'
The Tiger Global-led round values the startup at $1 billion as it wires stablecoin rails directly into a federally chartered bank, aiming to modernize the correspondent-banking plumbing behind cross-border payments....
KryptoKite aggregates and summarises third-party crypto news. This is informational content, not investment advice. KryptoKite does not recommend buying or selling any asset.