Prism 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
content ownership
Latest stance: user retains rights
“You are responsible for the code, scripts, files, datasets, packages, prompts, instructions, credentials, and other content you or the Hermes agent place in a Prism workspace, and for the actions the agent takes on your behalf. You represent that you have the rights needed to submit that content and to operate the resulting workloads. You retain ownership of your code and data. We process them only to operate the service, run the agent, execute the commands you authorize, deliver media, troubleshoot, prevent abuse, and meet legal obligations. We do not use the contents of your workspace or your agent transcripts to train foundation models. The Hermes agent and AI-generated output may be inaccurate, incomplete, offensive, similar to output generated for other users, or unsuitable for your intended use. The agent operates a real machine with sudo access, so commands it runs can delete files, install software, send network traffic, or otherwise change state in your workspace. You are responsible for reviewing what the agent does and what it produces before relying on it, publishing it, or pushing it to systems outside Prism.”
Open citationJul 20, 2026traininglow
model training
Latest stance: no training claim
“You are responsible for the code, scripts, files, datasets, packages, prompts, instructions, credentials, and other content you or the Hermes agent place in a Prism workspace, and for the actions the agent takes on your behalf. You represent that you have the rights needed to submit that content and to operate the resulting workloads. You retain ownership of your code and data. We process them only to operate the service, run the agent, execute the commands you authorize, deliver media, troubleshoot, prevent abuse, and meet legal obligations. We do not use the contents of your workspace or your agent transcripts to train foundation models. The Hermes agent and AI-generated output may be inaccurate, incomplete, offensive, similar to output generated for other users, or unsuitable for your intended use. The agent operates a real machine with sudo access, so commands it runs can delete files, install software, send network traffic, or otherwise change state in your workspace. You are responsible for reviewing what the agent does and what it produces before relying on it, publishing it, or pushing it to systems outside Prism.”
Open citationGenerated from live stance events. Informational only, not legal advice.