Avoiding Data Leaks: Unknown Perils of Online Translators

When you translate an important letter or personal message with an online translator, it all seems simple and convenient: you enter the text, press a button, and within moments, you have your translation.

But have you ever considered what happens to your data at that moment? Where is it stored, who can access it, and how secure are these services?

In this article, we’ll explore how your data is handled during machine translation and what risks might be involved.

How Machine Translation Works

Machine translation (MT) is the automatic translation of text using computer programs. Most modern MT systems, such as Lingvanex, DeepL, and Google Translate, use neural networks. This means they’re trained on vast amounts of data using neural network technology, mimicking how the human brain processes information.

For translated text to match the quality and meaning of the original, neural machine translation models are "fed" extensive amounts of texts, essentially acting as textbooks for training. The more comprehensive and high-quality the training data, the better the translation quality. This leads us to a critical issue: where can we obtain such a large quantity of varied, quality information?

Any publicly available information in multiple languages can serve as “training data” for language models — official international documents, news, literature, scientific materials, and even users’ personal data online.

Types of Data Transmitted in Translation

When using an online translator, you either paste text, upload a document, or enter text through voice input. We may do this without considering where the message from a foreign friend, corrections from an international client, or information from a bank consultant overseas might end up.

It’s worth thinking about. Few people pay attention to notifications about data collection, and online translation services' privacy policies often contain clauses like:

"This license allows the Company to:

Host, reproduce, distribute, communicate, and use your content — for instance, to save it on our systems and make it accessible anywhere you go.

Publish, publicly perform, or display your content if you've made it visible to others.

Modify and create derivative works based on your content, such as reformatting or translating it."

What Exactly Can Be Used with Our Consent

Let’s look at this in detail.

Input data includes any information entered by the user:

• Text documents. Plain text files, Word documents, PDFs, and other formats containing text for translation.
• Personal messages. Text from messengers, email, and social media.
• Web pages. HTML code and webpage content for site translations.
• Audio and video. Speech data for translating spoken content (when using speech recognition and translation systems).

Translation may also involve metadata accompanying the input data:

• Language tags. Information about the source and target languages.
• Structural data. Details about document formatting, such as headers, paragraphs, lists, and other elements.
• Contextual information. Data about the text's context, such as subject matter, style, and target audience.

Data Storage

Most online translators store translated data on their servers. This may involve short-term storage to ensure smooth service operation or long-term storage for improving translation quality and developing new features.

Some translated content may be temporarily cached in your browser for faster future translations of the same texts. Mobile apps might store translation history locally on your device, while popular online translators, like Google Translate, store translation history in the cloud, synchronizing it across your devices.

What Happens with This Information?

Translation service providers often use translation data to train algorithms and improve translation quality.

Some services may use your translation history to personalize results, offering the most relevant translation options based on previous queries.

In some cases, anonymized translation data may be used for marketing purposes, such as creating targeted ads.

Security Risks and Threats of Data Storage

According to a 2023 IBM and Ponemon Institute study, the average cost of a data breach has reached a record $4.45 million, making it too expensive to ignore.

In 2017, a data breach occurred at Statoil due to employees using a free online translation service for internal documentation. Information intended to remain private ended up being indexed by Google.

These examples show that data leaks don’t always result from hackers; user errors are often the cause. Employees may not always realize the consequences of their actions. Free online translators can create a false sense of security because they’re accessible without payment or registration.

Recommendations

Companies should avoid using unsecured, free services for handling sensitive information. Instead, they should opt for local servers within their corporate network or choose cloud services from trusted providers with robust security measures.

For example, the Lingvanex on-premise machine translation solution is software installed on company devices. This keeps all confidential information processed internally without leaving the organization. A local translator can operate offline, eliminating the risk of data leaks. Company management can set access rules and configure security parameters for these systems.

Conclusion

When working with your own or others' data, it’s essential to carefully assess the risks of each action. Yes, online translators offer a quick, cost-free way to translate documents, but a penny saved on translation could lead to significant losses from leaked confidential information.