The Right to Be Forgotten Does Not Delete Anything

The judgment in Google Spain SL v AEPD and Mario Costeja González (Case C-131/12, 13 May 2014) established that a search engine operator processes personal data in its own right, and is therefore a data controller with its own obligations, separate from the publishers whose pages it indexes. That was the structural move. From it followed the practical consequence: an individual can require a search engine to remove specific results returned for a search on their name, where the linked information is inadequate, irrelevant, no longer relevant, or excessive relative to the purpose of the processing, and where no overriding public interest applies.

The newspaper, La Vanguardia, kept its archive intact. The Court was explicit that the balance for a media archive differs from the balance for a search engine, because a search engine assembles a structured profile of a person from otherwise scattered material and makes it available to anyone who types a name.

What delisting actually does

A granted request removes a particular URL from the results shown when someone searches your name on that search engine. Everything else remains.

• The page stays published. The publisher is unaffected and keeps its own editorial rights.
• Other queries still find it. Search the article's headline, the employer, the case number, or the street, and the result reappears. Delisting is bound to name queries.
• Each search engine is separate. A grant from one is not a grant from another. Every operator has its own form and its own assessment.
• Archives are untouched. The Internet Archive, national library web archives, and third-party scrapers are outside the mechanism entirely.
• Publishers are usually told. Search engines commonly notify webmasters that a URL has been delisted. The requester is not named, but news organisations have repeatedly worked out who asked and written follow-up stories about the request itself.

That last point is the trap. A delisting request against a news article can produce a fresh article about the delisting request, which is itself indexable, and which frequently ranks well because it is recent. Anyone considering a request against journalism should weigh that outcome as a realistic possibility rather than an edge case. The same dynamic sits behind a lot of failed reputation cleanup attempts.

The territorial fight

France's data protection regulator, the CNIL, took the position that delisting had to apply globally, and fined Google in 2016 for limiting it to European domains. The Court of Justice resolved this in Google LLC v CNIL (Case C-507/17, 24 September 2019).

The ruling: EU law does not require delisting across all versions of a search engine worldwide. It requires delisting across all member state versions, combined with measures that seriously discourage access from within the EU to the delisted result on non-EU versions. In practice that means geo-blocking based on the searcher's location rather than the domain they typed.

The Court also noted that EU law does not prohibit a national authority from ordering global delisting in an appropriate case, which left the door ajar. The default, though, is regional.

How the balance is struck

Article 17 of the GDPR now codifies a right to erasure, and Article 17(3) lists the exemptions: freedom of expression and information, compliance with a legal obligation, tasks in the public interest, public health, archiving and research, and the establishment or defence of legal claims. The freedom of expression exemption does most of the work in delisting cases.

A companion 2019 ruling, GC and Others v CNIL (Case C-136/17), addressed results containing special category data such as political opinions, religious belief, health, or criminal offences. The Court held that these attract stronger protection, but that a search engine must still weigh the public's interest in access, taking into account the nature of the information, its sensitivity, and the applicant's role in public life. For criminal proceedings, it said the passage of time matters: a link to coverage of a since-resolved or superseded stage of proceedings may need to be delisted even where coverage of the outcome does not.

The English High Court applied similar reasoning in NT1 and NT2 v Google LLC in 2018, two claims heard together, both from businessmen with old convictions. One succeeded and one did not. The distinguishing factors included how long ago the conviction was, whether the individual continued in a comparable business role, whether they had shown remorse, and whether they had misled the court. Rehabilitation is treated as a fact to be assessed rather than a status that arrives on a date.

Factor	Favours delisting	Favours keeping the result
Role of the person	Private individual	Public figure, politician, executive
Age of the information	Old, superseded	Recent, ongoing
Accuracy	Inaccurate or outdated	Accurate and current
Relation to professional life	Purely personal matter	Relevant to how they do business now
Source	Forum post, scraped aggregator	Journalism, official record

Google's own transparency reporting on the mechanism describes millions of URLs evaluated since 2014, with slightly under half of them delisted. The largest single category of requesters is private individuals asking about personal information on directories, social profiles and aggregator sites, not people contesting news coverage.

Outside Europe

There is no general equivalent in the United States, where the First Amendment makes a right to suppress accurate published information about a person very difficult to construct. California's narrower instruments address different things: a law allowing minors to remove their own posted content, and the Delete Act's centralised deletion request mechanism aimed at registered data brokers rather than at publishers or search engines. That broker channel is closer in spirit to the broker opt-out process than to European delisting.

Elsewhere the picture diverges sharply. Several jurisdictions have adopted comparable rights by statute or judicial decision. Brazil's Supreme Federal Court went the other way in 2021, holding that a general right to be forgotten is incompatible with the Brazilian Constitution. Twelve years after the Costeja ruling, the jurisdictions are not converging.

If you are considering a request

1. Start with the source. If the page belongs to a data broker, a people-search site, or a defunct forum, ask the operator to remove it. Removal at source solves the problem everywhere, permanently, and does not require anyone's balancing test.
2. File per search engine, per URL. Requests are evaluated one URL at a time. A vague request covering "everything about me" gets a poor response.
3. Explain the harm concretely. Why the information is inaccurate, outdated, or irrelevant to any current public interest, in specifics. Assessors are weighing named factors, so address them.
4. Expect the publisher to be notified. Decide in advance whether you can live with that, particularly for a news source.
5. Escalate to your data protection authority if refused. The search engine's decision is not final, and national regulators take complaints.
6. Remember what remains reachable. Delisting affects a name search on one engine in one region. It does not touch archives, non-name queries, or anyone who already saved the page.

The mechanism is genuinely useful for the case it was built for: an ordinary person whose worst week a decade ago is the first thing a stranger sees when they type a name. It was never designed to remove information from the world, and treating it as though it were leads people to file requests that make their situation worse. The realistic goal is to break the link between your name and a specific page in one place where most people look, which is a real improvement and a limited one.