kapynAI / Models

Disrupting a coordinated model-distillation campaign

OpenAI blocks a coordinated model‑distillation attack that sought to extract protected reasoning. The company identifies the campaign, disrupts the extraction attempts, and strengthens defenses against adversarial distillation, highlighting the growing threat of model extraction. This move underscores the importance of robust security for large language models.

OpenAI Blog·Sep 30, 2026

Opening Kapyn…