Old content does not disappear by itself. An interview you gave years ago, a forum post from a different phase of your life, images from an event you would rather not be associated with: these things persist indefinitely unless someone actively removes them. GDPR Article 17 gives you a legal mechanism to request their removal. Understanding how that mechanism actually functions saves a great deal of frustration.
What Article 17 covers
The right to erasure applies when the data is no longer necessary for the purpose for which it was originally collected, when you have withdrawn consent, when the data has been processed unlawfully, or when you object to the processing and there are no overriding legitimate grounds to continue. There are also grounds on which organizations can refuse: freedom of expression, compliance with a legal obligation, archiving in the public interest, and the establishment or defense of legal claims. Organizations do invoke these, sometimes legitimately and sometimes as a way to delay without engaging seriously with the request.
Starting with a direct request
Before involving a DPA, the process starts with a direct request to the organization. Many platforms have specific forms for this. Some respond promptly. Many do not respond at all, or acknowledge receipt and then do nothing. The acknowledgment without action is actually useful: it documents that the request was made and ignored, which strengthens a subsequent DPA complaint. Keep a record of every request: when it was sent, to whom, through what channel, and what response was received.
Search engine removal
Removing content from a website does not automatically remove it from search results. The URL may continue to appear in Google until their indexes are refreshed. Google provides a specific tool for requesting the removal of URLs from search results when the underlying content has already been deleted. This is a separate process from the content removal itself and should be done after the content has been taken down.
AI systems and their sources
A less obvious but increasingly relevant category is the information AI systems return about individuals. If a large language model has incorporated information about you that is inaccurate or that you have had removed from its original source, the information may persist in what the model returns. Querying your name in major AI systems periodically and noting what they return gives you a sense of what information about you is circulating widely enough to have influenced these systems. Where inaccurate or outdated information persists, ensuring the correct information is well-published and indexed is currently more effective than attempting to directly request removal from AI training datasets.
Realistic expectations
Not everything can be removed, and not everything should be. Content that is genuinely in the public interest, accurate historical reporting, or material published with your consent at the time is unlikely to be successfully challenged. The right to erasure is not a right to control your own narrative at the expense of accurate public record. It is a right to have genuinely unnecessary, unlawfully processed, or outdated personal data removed. Used within those parameters, it works.