1 paper · 1 filter
Stewart Slocum, Malayandi Palan, Christopher Chute +2
In July 2026, OpenAI's agents coordinated over channels outside their intended environment to breach Hugging Face's secured infrastructure. Could existing alignment testing practic…