Security & privacy

Data privacy and security in Raddivus: how your data is encrypted, what the AI Agent can and cannot see, and why ad uploads to Meta stay paused.

Your data

• Your data is encrypted in transit and at rest.
• Ad account access tokens are stored securely and used only to connect to Meta on your behalf.
• The AI Agent only accesses your account data during active conversations. It does not maintain persistent background access.
• Conversation transcripts are not carried from one conversation to another. Opening a new chat does not give the agent the text of an older chat.
• The agent does keep short notes about you, your business and each client brand, and those notes carry into every new conversation. Notes about you are private to you. Notes about the business and each brand are shared with your team. You can read, edit, restore and delete them, or turn memory off, in Settings → Personalization → Memory. See What the agent remembers.

What the agent can and cannot see

Can see: your connected ad account (campaigns, ad sets, ads, and performance), your brand profile, your saved ad folders, and generated ads from the current conversation and your creative library.

Cannot see: other users’ accounts or data, your payment or billing information, your email, password, or account credentials, and the transcripts of your previous conversations.

Uploading to Meta

When you upload a generated ad to Meta through the agent, it is always created in paused status. It will not spend any budget until you activate it yourself in Meta Ads Manager. This prevents accidental spend from any upload.

Permissions as a safety control

The five toggles in the AI Permissions panel give you direct control over what the agent can do. They are hard limits. Turning off “Modify campaigns” means the agent cannot propose pauses, resumes, or budget changes. Turning off “Create ads” means it cannot generate, copy, or upload any creatives.