Kilvin policy evolution
Before/after stance changes across captured policy versions, with exact citations. If no before/after delta is available yet, AIRIN shows the latest citation-backed stance events instead.
No before/after stance delta is available for this filter yet. Latest citation-backed stance events are shown below.
Jul 20, 2026traininglow
model training
Latest stance: no training claim
“AI features in the Service — including AI-assisted development and any AI capabilities in software we deliver — are served through Kilvin's own model gateway. All model traffic exits our environment through this single controlled path to three upstream providers: Anthropic — direct API access to Claude models. Anthropic does not use API inputs or outputs to train models.”
Open citationJul 20, 2026traininglow
model training
Latest stance: no training claim
“AWS Bedrock — model inference within AWS. Amazon does not use Bedrock inputs or outputs to train models.”
Open citationJul 20, 2026traininglow
model training
Latest stance: no training claim
“Data submitted to AI models through the Service is not used to train models, subject to the upstream providers' terms. Our model gateway does not retain prompt or response content; it records only the usage metadata described in Section 1. Separately, when you interact with AI features that maintain a conversation — such as AI agent sessions — the conversation transcript is stored within your organization's workspace as part of the Service, so you can review and resume your sessions. Transcripts are subject to the same access controls, encryption, retention, and deletion terms as your other Client Data.”
Open citationGenerated from live stance events. Informational only, not legal advice.