Responsible Deployment Policy

This policy applies to every model, feature and product ACATHRCA makes available to the public, including ChatGiZa on the web and ChatGiZa Agent for VS Code.

• Every significant change is evaluated for quality, factual accuracy and misuse risk.
• Features that can take actions, such as agents, are tested for unintended or unsafe behaviour.
• Features that fall short of our bar are delayed, limited or not released.

We release new capabilities gradually where we can, starting with smaller groups of users, so that problems surface before they reach everyone.

• We monitor for abuse and unexpected behaviour, with privacy protections in place.
• We can restrict or roll back a feature quickly when the evidence calls for it.
• We review incidents and update our evaluations so the same failure is caught earlier next time.

Anyone, inside or outside ACATHRCA, can report a concern about how we deploy AI to security@wellxai.world. We do not retaliate against people who raise concerns in good faith.