Machine translation: when technology puts confidentiality at risk

The birth of machine translation

It is important to remember how we got where we are today. Machine translation started off as a futuristic idea and an attempt to imitate humans’ ability to understand and translate languages. Initial efforts were based on simple sets of rules, but as machine learning and neural networks emerged, MT underwent a radical transformation. Tools like Google Translate and DeepL, which facilitate communication on a global scale, are just some of the success stories in this sphere.

All kinds of data involved

To really grasp these confidentiality issues, looking at the nature of the information being translated is key:

• Business: From financial documents to strategic plans, businesses regularly use MT for international dealings.
• Personal: In a connected world, MT is often used in personal interactions with family, friends and acquaintances who speak different languages.
• Research: Researchers use MT to access studies in other languages, thus broadening their frame of reference.
• Medical: Extremely sensitive medical files, reports and diagnoses are often translated for international use or consultations.

MT: a double-edged sword

Though machine translation offers incredible accessibility and is highly convenient, it poses some serious problems:

• Storage and retention: Where are the data involved in the translation process kept? For how long are they kept? Can this information be accessed by third parties?
• Access and security breaches: Are MT systems – especially cloud-based ones – protected from cyberattacks?
• Use of data: Does the translated information contribute towards improving MT systems? Is it used for commercial purposes, i.e. sold to third parties?

Translation agencies: back to basics?

Given these concerns, many are turning to traditional translation agencies to fulfil their translation needs. These businesses collaborate with professional human translators, which means that, as well as high-quality, nuanced translations, they offer a level of confidentiality machines cannot match.

The impact of regulations on MT

With the introduction of laws like the GDPR in Europe, data protection has become a central issue. These regulations require MT providers to guarantee that users’ information will remain confidential. But the question remains: are they applied uniformly across the board? And what about services based outside of the GDPR’s jurisdiction?

Strategies to use MT securely

When using machine translation, some preventive measures are required. These include:

• Choosing carefully: Not all MT services are created equal. Some guarantee more confidentiality than others.
• Local translation: Choose tools that work locally, on your device, without sending data via the internet.
• Strengthening security: Using VPN, anti-malware software and other security measures is crucial if you want to stay protected.
• Avoid translating sensitive information: If possible, avoid translating extremely sensitive information using online MT services.

Practical cases: the implications of MT

• Business: A business looking to set up an international partnership translates its contract via an online tool. Sensitive commercial clauses might be exposed.
• Individuals: Someone translating a personal letter might unwillingly disclose information on their health, private life or finances.
• Academia: A researcher translating an unpublished discovery could risk having crucial information fall into the wrong hands.