translation

Security in Patent Translation

Patent translation entrusts us with information of the highest sensitivity — unpublished, pre-filing details of an invention. ip-pro treats the confidentiality, integrity, and availability of the information in our care with the same weight as quality, and builds that commitment into how we work. As AI and machine translation become more widely used, we define internally and in writing what may be used and how far, and put those rules into daily practice.

Handling of translation data

The translation data entrusted to us goes to the core of a client's business. We treat it as information held in trust, and we hold ourselves to the following.

• Translation data is managed separately for each client
• We do not reuse it for other clients' projects
• We do not use it for any purpose other than the one intended — including providing it to outside parties for machine learning
• Access to the information in our care is limited to the people who need it — the staff assigned to the project and the administrators — and every access is logged
• We accommodate clients' requests on how data is retained or disposed of once a project ends

Oversight and operation of internal systems

We oversee, design, build, and operate ip-pro's business systems and information infrastructure in-house — from cloud services to on-premises environments to employees' work devices. This means we always know, ourselves, exactly where and how a client's information is stored and processed.

• Employees' work devices are provisioned, configured, and managed in-house, and we monitor how they are used on an ongoing basis
• Access rights to servers and internal systems are granted only as far as the work requires, and we manage accounts centrally
• For the cloud services we use, we select them, configure them, and govern their scope of use ourselves
• We do not outsource the design, build, or operation of our systems — all of it stays in-house, end to end

Security and governance

In patent translation, having everyone judge by the same internal standards matters as much as the technical safeguards. We set those standards out in writing and build them into our daily work.

• We take on work on the basis of a non-disclosure agreement (NDA) with the client
• For every app, piece of software, and cloud service we use internally, we determine individually whether it may handle confidential or personal information
• For each kind of task — translation, proofreading, document comparison, internal sharing — we set out in writing which tools may be used and which may not
• We make sure the definitions and handling rules for confidential and personal information are understood across the team, so that everyone judges by the same criteria
• Only approved software and services may be used for work, and anything new goes through a request-and-approval process
• We review these policies and criteria on an ongoing basis as technology and our working environment change

Policy on the use of AI and machine translation

AI and machine translation hold great potential for both the quality and the efficiency of translation, but the handling of input data calls for care. With protecting the client's information as our first priority, we work to the following policy.

• Before using any AI tool, we explain its purpose and scope to the client and proceed only once they have agreed — never without consent
• Even then, we use machine translation and AI only on plans and settings under which input data is not used for training
• We avoid any service whose terms allow data to be retained or used for training — free services being the typical example
• Before adopting any external service, we check its data-use terms, whether data is retained, where the data is held, and so on, and decide on that basis
• Before adopting a new AI service, we evaluate it internally against the criteria above and decide accordingly

Operations that support quality

In patent translation, security and quality are two sides of the same coin. We keep the quality-check results and observations from each project on a per-client basis and draw on them to improve that same client's next project. This lets us reflect each client's preferences in style, terminology, and expression, project after project.