A marketing analytics firm replaces every customer name in its dataset with a randomly generated token and stores the lookup table that links each token back to a name on a separate, access-controlled server it still owns. It tells a client that, because the working dataset now contains no names, that dataset is anonymous and falls outside the GDPR. Why is this conclusion incorrect?
- AThe dataset is anonymous only if the lookup table is held by a third party rather than the firm itself, so ownership of the table is the deciding factor.
- BThe dataset is anonymous because tokenisation is a recognised anonymisation technique under Article 4, so no further conditions need to be met.
- CThe dataset becomes anonymous as soon as the firm contractually promises not to consult the lookup table, since the risk of re-identification is then removed.
- DBecause the firm retains a key that can re-attribute the tokens to identified individuals, the data remains pseudonymous personal data and stays within the GDPR. Correct
Why A is wrong: It is tempting because separation of the key from the data is a genuine pseudonymisation safeguard, but who holds the key does not convert pseudonymous data into anonymous data; identifiability by means reasonably likely to be used is what matters.
Why B is wrong: This is a plausible confusion of two distinct concepts, but Article 4(5) treats tokenisation that retains a re-identification key as pseudonymisation, not anonymisation, so the GDPR continues to apply.
Why C is wrong: A promise not to use the key is appealing as a safeguard, but anonymisation depends on whether re-identification is reasonably possible, not on a policy commitment; the key still exists and can be used, so the data stays personal.
Why D is correct: Correct: Article 4(5) defines pseudonymisation as processing where data can still be attributed to an individual using additional information kept separately, and Recital 26 confirms pseudonymous data is personal data because re-identification remains reasonably possible.