But the more tools we use, the more doors we can inadvertently open. All it takes is a poorly configured browser, outdated software, or a “free” tool that turns your translations into training material for artificial intelligence.
To prevent this from happening, we have prepared a quick overview of the most important elements that a translator should have under their control today.
1. Regularly Updated Operating System
It seems like an insignificant detail, but regular updates to your OS not only provide new features or perhaps faster startup times, but also increase security.
Developers use them to fix bugs that could be exploited by hackers. This can allow attackers to access your files, install malware, or track what you do without your knowledge. All it takes is one weak link and sensitive data can be leaked to anyone.
Recommendation: Turn on automatic updates. Don’t put them off indefinitely. Attackers often exploit known and unpatched vulnerabilities.
2. Combining Antivirus with Firewall = Basic Protection
Without this security duo, a computer used for translation can be like a wide-open door with a sign saying “Welcome Everyone”.
Antivirus protects you from common malware and ransomware, which can lock and encrypt all your data and demand a ransom. These little monsters can enter your system with ease: through a phishing email, an attached file from a “client”, the link you shouldn’t have clicked on, or even just by loading an untrusted website.
A firewall, in turn, monitors traffic – data that comes in and goes out. It monitors which apps are trying to connect to the internet and can block suspicious activity before you even notice it.
Recommendation: If you use Windows, the built-in Defender and firewall are highly reliable today. However, for greater control, you can also use third-party solutions such as ESET, Bitdefender or Avast.
3. Password Manager – Let Passwords Not Become Your Achilles’ Heel
Do you always use the same password with only a small variation? Or do you store your passwords in Excel? Strong and original passwords are the foundation of security, but remembering dozens or more is unrealistic.
This is what password managers are for – applications or browser extensions that generate, securely store, and autofill your passwords.
Recommendation: Try Bitwarden, 1Password, or the built-in password managers in Chrome or Firefox. Don’t forget to secure access to the admin console with a master password or biometrics (fingerprint, FaceID).
4. Two-Factor Authentication (2FA): Protection When Passwords Aren’t Enough
Even if you use long, complex passwords full of special characters, numbers and letters, there is still a risk that they can be leaked. That’s why it’s a good idea to add another layer of protection – two-factor authentication (2FA).
This means that after entering your password, you must still confirm your identity with a code sent either via text message or from an app like Google Authenticator, Aegis or Microsoft Authenticator.
Recommendation: Activate 2FA at least for your most important accounts – email, password manager, cloud services (Dropbox, Google Drive), and tools you use to work with translations.
Is it truly free? What hides behind free tools
Some translation tools will entice you with a free version. Sounds great – who wouldn’t want to save money? But it’s important to remember that “free” often doesn’t mean without a price. Only instead of money, you end up paying with something else: your data.
Why do companies offer free tools at all?
- They want to get as many users as possible for their business.
- They collect data about how you use the tool, what you translate, and what your habits are.
- Based on this, they can better target advertising, test new features, or even sell the collected data.
- And some companies use this data to train their AI models so that their tools can improve and learn from your texts.
➡️ In practice, this means that the text you enter into a “free” online translator, grammar checker or CAT tool can be stored, analysed and processed. This can happen even without your explicit consent. It’s simply enough that you used the tool once.
If you are working with confidential documents, medical reports, contracts or legal texts, we recommend using a paid and secure tool or a local (offline) version of the software.
Four questions to check if the tool you’re using plays fair and what to look out for before using it
Not every tool is automatically bad. You just have to put in the effort and find out how it handles your data. Look for answers to these key questions in privacy policies (yes, those long documents that most of us skip):
1. Where is your data stored?
Good tools will:
- Clearly state which country their servers are in; ideally the EU, Switzerland, or other countries with strong privacy protections.
- Affirm that the data is encrypted and will not be used for other purposes.
- Outline their privacy policy in simple and clear terms without obscuring the facts.
Problematic tools will:
- Not mention server location at all, or use phrases such as “in accordance with applicable laws” (without specifying which ones).
- Host servers in countries where privacy protection is weaker or legislation allows access to data by state authorities (e.g. USA, China, India).
- Say nothing about what happens to your data after it is uploaded.
2. Does the tool have any security certifications?
This is not just a formality. Certifications like ISO 27001 or SOC2 mean that the business has transparent processes in place to protect data. These ensure that your online communication is not left to chance.
Bonus: If the tool is GDPR compliant, this should be clearly stated. It means that you have the right to know what is happening with your data and to have it deleted at any time.
3. Is your data used to train artificial intelligence?
Many modern tools today claim to “improve user experience”. In translation, this often means that your personal data ends up in AI training datasets. If the privacy policy says something like:
- “We may process your data anonymously to improve our services.”
- “Content may be used for analytical or development purposes.”
… this might actually mean that your translation project becomes a data source for a neural network.
If you are translating something that is not intended for public use, be highly cautious. Your data is more valuable than you might think.
4. Is data transfer and storage secured by encryption?
Your text should be protected both in transfer and at storage. This includes:
- HTTPS – when you enter text in the web interface, the connection must be encrypted.
- AES-256 or RSA – encryption standards ensuring that even if someone were to gain access to your data, they would not be able to read it.
If the tool you’re using doesn’t mention encryption at all, this is a red flag. Trustworthy providers tend to communicate this clearly and directly because it’s a badge of honour.
Different tools, different purposes: What to watch out for when working with PDF, machine translation and AI
As the tools we use at work become more convenient, the amount of data transferred to these services grows as well. And not all of them handle your data equally responsibly.
See an overview of selected tools and platforms that looks not at features, but at how they approach security, encryption, data storage, and their further use, especially when it comes to training AI models.
Our goal is not to nudge you to use specific tools. However, when working with confidential or sensitive materials, we recommend also looking at what the tool knows about you and what it can store.
OCR tools (Optical Character Recognition – Extracting text from PDF and images)
SmallPDF
- Data is stored on their servers temporarily
- They do not list any safety certificates
- They use SSL/TLS encryption
- They don’t use your data to train AI models
✅ Pros: Easy to use, works with cloud services (Dropbox, Google Drive)
❌ Cons: Free version is limited, questionable privacy practices
ILovePDF
- Data is only stored temporarily, automatically deleted within 2 hours
- It is ISO/IEC 27001 certified
- Encryption via HTTPS
- Data is not used to train AI
✅ Pros: Comprehensive tools for working with PDF, very user-friendly interface
❌ Cons: Data may be processed on third-party servers
Machine Translation (MT)
Google Translate
- Data can be stored indefinitely
- ISO 27001 certified
- Encryption via SSL/TLS
- Your data is used to train AI models, and there is no way to opt out of this process
✅ Pros: Huge language range, regular updates
❌ Cons: Your translations can be used to improve the service
Microsoft Translator
- Data can be stored indefinitely
- ISO 27001 certified
- Encryption via SSL/TLS
- For regular users (B2C), data is used for training
- For corporate customers (B2B), training is turned off by default – it can be set up privately
✅ Pros: Good integration with MS Office, stable service
❌ Cons: Data can be used for training, depending on the version
DeepL
- Data is stored temporarily
- ISO 27001 certified
- Encryption via SSL/TLS
- In the free version, data is used for training
- In paid versions, there is an option to opt out from model training
✅ Pros: High-quality translations, clearly set privacy policy
❌ Cons: Free version has limits
🤖 AI tools (generative artificial intelligence)
ChatGPT
- Data can be stored indefinitely
- It does not list any safety certificates
- Encryption via SSL/TLS
- Your data is used for training, but you can opt out of this process in the settings
✅ Pros: Great flexibility, fast response, continuous improvement.
❌ Cons: Your prompts will be used for training, unless you manually disable this in the settings (incognito mode).
Claude, Gemini, LeChat, Perplexity
- Data storage varies: Claude, for example, keeps data for 30 days, other platforms do not provide exact information
- Certifications mostly not listed
- Encryption via SSL/TLS
- Most tools use data to train models, but in some cases an opt-out setting is available
✅ Pros: Advanced AI features, multiple usage modes.
❌ Cons: Similar risks to other AIs, especially from a privacy perspective.
Privacy is not an afterthought. And especially not when it comes to translations.
Translation tools are constantly improving, but so is the amount of data we share with them. Not all services approach privacy protection the same way, and they can take more from your work than just the text. Here are the most common pitfalls to watch out for and how to elegantly avoid them.
Data collection and further use
For numerous online tools, you are not only a user, but also a valuable source of data. The services track what you do, for how long, what text you enter, and then use it to improve services and develop new features, or simply for statistics. Sometimes even storing all this info in their artificial intelligence training database.
How to opt out of this?
- Look for tools with a clearly outlined privacy policy.
- Look for a data collection opt-out option.
- Be sure to read what specific data the services collect, and especially what they use it for.
Where is your data stored and how is it protected?
The texts and data you send to these tools are often stored on remote servers. Without sufficient encryption, there is a risk that they could be misused, whether through a leak, a system error or unauthorised access.
What to watch out for?
- Make sure that the service uses end-to-end encryption
- It is important that data is encrypted during transfer and when it is stored
- It is a good sign when data is stored in countries with strong privacy legislation (e.g. the EU)
Do the services train artificial intelligence with your translations?
Your texts can help make AI smarter. The question is – did you consent to it? Some tools will turn your work into their very own textbook. Even when you are translating private correspondence, legal documents, or something that only your client should be able to see.
What to watch out for:
- Serious tools will give you the option to switch this off.
- Always look at whether and how data is used to train models.
- If this is not clear, just assume the services use your data.
Third parties: Who else has access to your data?
Occasionally, a tool might look secure, but in the background it collaborates with other companies that process your data. And they may abide by entirely different rules.
The solution?
- Look for tools that process everything locally or have clearly disclosed terms and conditions for their partners.
- If there is no mention of a third party, there either is none or they would rather not mention it. Neither option is ideal.
Common sense and “rules of thumb” in digital security
You don’t have to be a cybersecurity expert to make smart decisions. Just a few simple principles will help you uncover weaknesses in the tools you might be using every day.
Well-known brands vs. no-name services
With well-known tools, you have a better chance that they know what they are doing and are accountable for how they do it.
Well-known tools
✅ They have a reputation they want to protect
✅ They have likely passed security audits
✅ They have transparent terms and conditions and a dedicated team that deals with data protection
Unknown (or little-known) services
⚠️ They may seem “fast and free”, but often lack information about where your data goes
⚠️ They do not disclose security standards or where your data is stored
⚠️ They do not offer the option to change privacy settings – because they have none
Litmus test? If the tool has no privacy page, or their policy is full of empty phrases, it’s time to close the window.
Free account, registration, paid version – what you gain (or lose)
Not all accounts are created the same. Once again, more control usually comes with a price.
Free version
- Minimum features, minimum control
- Your data can be used almost without restrictions
- Often no option to turn off data collection or AI training
Registered account (free)
- You get access to more features
- Sometimes basic privacy preferences can be managed
- However, you are still a “product”, not a customer
Paid account
- Access to full functionality
- Option to turn off data collection or AI training
- Much more control over what happens to your content
Obscure privacy policy? Ask AI about it
Not sure what the phrase “data may be processed in accordance with internal policy” means? Copy it into ChatGPT (or a different AI tool) and ask it directly:
“Explain this part of the privacy policy to me in simple terms.”
This will help you quickly gain an overview, and you may even learn that some tools are better not used at all.
These rules will help you use online tools without having to deal with many inconveniences.
