OpenAI announced that it has disrupted a coordinated campaign aimed at extracting protected model reasoning from its systems, according to a company disclosure.
The company described the operation as an attempt at adversarial distillation — a practice in which third parties try to replicate or extract capabilities from a protected model. OpenAI said the campaign targeted protected reasoning specifically.
In response, OpenAI said it is strengthening its defenses against adversarial distillation. The disclosure did not include details on the scale of the campaign, the actors involved, or the specific techniques used.
The announcement underscores growing industry concern over efforts to copy or extract value from proprietary AI models without authorization.