Privatemode is a Confidential AI service whose core goal is to protect data security and privacy while using AI, achieving end-to-end encryption.
In terms of use cases, Privatemode is aimed at organizations that need to process sensitive data in AI workflows. Typical scenarios mentioned in the materials include: automated claims processing, accelerating government services, automating legal workflows, and shortening document processing time. Applicable industries include life and health insurers, public institutions, law firms, legal tech providers, and health tech companies. These scenarios involve sensitive content such as claimants' medical records, citizens' personal information, clients' confidential documents, and patients' health records.
In terms of key capabilities, Privatemode is based on Confidential Computing. Data is encrypted before it leaves the user's device and remains protected throughout AI processing, avoiding plaintext exposure. It provides end-to-end confidentiality, with runtime encryption also protecting data during data processing; provides end-to-end verifiability, with hardware-based certificates that can prove the integrity of the Privatemode backend; and has external audits confirming these security properties. The materials mention that Privatemode runs on open frontier AI models and received the Best of Technology Award for Privatemode AI.
In terms of how to use it, the materials say "Confidential AI, only a click away," and suggest watching a live demonstration of Privatemode. Specific registration, deployment, onboarding steps, and installation commands are not yet provided.
In terms of compliance and security certifications, the materials state that Privatemode meets the German Federal Cyber Security Agency's "very strong attestation" BSI C5:2026 standard for confidential computing in the cloud; has been reviewed by a leading consulting firm and is recommended for professions handling sensitive data such as doctors, lawyers, and tax advisors (StGB §203 ready); has had its security and privacy properties confirmed through review and audit by a Big Four accounting firm and a leading cybersecurity consulting firm; through open source, reproducible builds, transparency logs, and end-to-end remote attestation, enables anyone to verify the integrity of Privatemode; and is additionally certified under ISO 27001. The FAQ states that Privatemode has undergone two independent third-party security assessments: one focused on penetration testing, and the otherfocused on architecture review. To date, all audits have confirmed the validity of its security properties, and audit reports can be provided upon request by enterprise customers.
Compared with traditional AI services, the materials point out that prompts and data for powerful AIs such as ChatGPT and Claude are processed in plaintext on the provider's servers, with no verifiable privacy guarantees; while self-hosted solutions can provide full control over data, they require dedicated hardware, in-house MLOps, and ongoing maintenance, with timelines of weeks to months, high upfront and recurring costs, and full responsibility borne by the user. Privatemode is positioned as "cloud-grade AI without compromising on privacy," enforcing compliance through technology rather than contracts.
Regarding specific technical mechanisms such as whether AI learns from or memorizes user data, and how AI processes data under "always encrypted" conditions, the materials do not elaborate; specific pricing, model lists, and usage restrictions are not yet provided.
