Does pseudonymised data stay under GDPR?

Yes, as long as a key exists somewhere that can restore the link. But GDPR explicitly names pseudonymisation as an appropriate safeguard. The personal data you replaced is then no longer in what you send to an AI vendor.

Is anonymisation always better?

No, it is more final, not better. For AI work pseudonymisation is usually more useful: the AI's answer only becomes usable once you can put the real names back.

Who can access my recovery key?

The recovery key (Identity Key) is given to you at the end of the run. ShareSafe Free keeps no copy, so keep it somewhere safe.

Does permanent replacement make my document anonymous?

Not by itself. There is no key, so nobody can restore the replaced values. But other details in the text, such as a role, a date or a place, can still point to someone. Check those during review if the document has to be anonymous.

Can an anonymised document become identifiable again later?

That risk exists when a lot of context remains. That is why anonymisation also deals with indirect identifiers, and why the review step stays part of the method.

All guides

Pseudonymisation or anonymisation: the difference, and when to choose which

The short answer. Pseudonymisation replaces personal data with placeholders while a key exists that can restore the originals. Anonymisation means nobody can reasonably identify the person any more: no key, and no other details that point to them. Legally the difference is large: pseudonymised data remains personal data under GDPR, truly anonymised data falls outside it. In ShareSafe, you choose reversible replacement when you work with AI and want to put the real values back into the answer, and permanent replacement when there should be no way back. Permanent replacement removes the key, but whether the result is anonymous depends on what else is left in the text.

→ Choose how to replace

The difference in one table

	Pseudonymisation	Anonymisation
What happens	Values become consistent placeholders	Values are removed or generalised so nobody can be identified
Reversible	Yes, with the key	No, for anyone
GDPR status	Remains personal data (Art. 4(5))	Outside GDPR (Recital 26), if truly irreversible
Recognised as	Security measure (Art. 32)	Data that is no longer personal data
Typical use	AI work, analysis, internal reuse	Publication, archive, external sharing
In ShareSafe	Reversible replacement: you get the recovery key (Identity Key), ShareSafe Free keeps no copy	Permanent replacement: no key is created. Whether the result is anonymous depends on the rest of the text

The trap: "anonymised" data that isn't

GDPR sets the bar high: nobody should reasonably be able to re-identify the person, including by combining other sources. Much of what gets called "anonymised" in practice is not:

initials instead of names (J.d.V. from a named small town is identifiable);
removing the name but keeping role, company and date;
black bars with the text still underneath;
a replacement table sitting in someone's mailbox. That is pseudonymisation with a badly managed key, the worst of both worlds.

The same goes for a document without a key: if the role, the company and the date are still there, someone can still be identified. Placeholders replace personal data; contract terms, trade secrets or medical content stay readable. In the review step you can select extra text and have it replaced too.

Be honest in your own documentation about which of the two you are doing. That is why ShareSafe names its modes after what they do: reversible replacement with the key in your hands, or permanent replacement with no key at all.

When to choose which

Choose reversible replacement (pseudonymisation) when:

you are giving the document to ChatGPT, Claude or another AI and want to put the real names back in the answer;
you analyse internally but need the link later;
you need consistency: [NAME_01] must stay the same person throughout the document.

Choose permanent replacement when:

the document goes outside: publication, sample case, training material, benchmark;
your retention obligation has ended but you want to keep the content;
you want certainty that even you can no longer restore the replaced values.

If the goal is a truly anonymous document, also remove or generalise the indirect details (roles, dates, places) during review.

How it works at ShareSafe.ai

Drop your file into the tool and pick your mode.
Check the before and after view; add anything detection missed.
Download the protected file. With reversible replacement you also get the recovery key; ShareSafe Free keeps no copy, so if you lose it, we can't put the values back either.

FAQ

Does pseudonymised data stay under GDPR?: Yes, as long as a key exists somewhere that can restore the link. But GDPR explicitly names pseudonymisation as an appropriate safeguard. The personal data you replaced is then no longer in what you send to an AI vendor.
Is anonymisation always better?: No, it is more final, not better. For AI work pseudonymisation is usually more useful: the AI's answer only becomes usable once you can put the real names back.
Who can access my recovery key?: The recovery key (Identity Key) is given to you at the end of the run. ShareSafe Free keeps no copy, so keep it somewhere safe.
Does permanent replacement make my document anonymous?: Not by itself. There is no key, so nobody can restore the replaced values. But other details in the text, such as a role, a date or a place, can still point to someone. Check those during review if the document has to be anonymous.
Can an anonymised document become identifiable again later?: That risk exists when a lot of context remains. That is why anonymisation also deals with indirect identifiers, and why the review step stays part of the method.

ShareSafe.ai keeps raw files in the EU. Minimal retention. You hold the key. Try it with your own document →