security· Models
Internal checks now require providers that don't train on or keep your data
The checks we run on your messages, such as the prompt-injection screen, now go only to model providers that don't train on what we send and keep nothing after they answer.
The model behind these checks was already served by such a provider. What's new is that every request now requires it, so a change of model or provider can't quietly send your text somewhere that keeps it. If no provider qualifies, the check fails instead of falling back, and a failed check behaves as it always has: the prompt-injection screen lets the run go ahead, as the security page says. Change summaries and repository descriptions are generated text, not decisions, and still run under the provider's standard terms, which allow keeping data for a limited time.