How Do EU Data Protection Authorities Explain the Right to Be Forgotten for News Articles?

The Right to Be Forgotten (RTBF) is often described as if it were a single removal mechanism. In practice, the official guidance reviewed in this study describes a more limited and more procedural reality. A person may ask a search engine to stop showing specified links for a name-based search, while the underlying article can remain online at the publisher. Whether delisting is justified can depend on factors such as freedom of expression, public interest, the age and accuracy of the information, the individual's public role and the nature of the underlying event.

Info Remover reviewed official data protection authority guidance, FAQs, complaint procedures and directly relevant decisions across 12 EU countries: Germany, France, Spain, Italy, the Netherlands, Belgium, Ireland, Austria, Sweden, Croatia, Malta and Poland. The fixed study contained 96 coded country-criterion fields (12 countries x 8 criteria). The final publication dataset contains 71 Yes, 13 Partially, 0 No clear guidance found and 12 Unknown / inaccessible, giving 84/96 (87.5%) substantively assessed coverage.

The strongest recurring cross-country pattern in the coded documentation is procedural. All 12 countries have a verified complaint or escalation route in the material reviewed. Substantive balancing factors are less uniformly documented. Journalism or freedom-of-expression considerations are clearly documented in 10 of 11 assessed countries, with one additional country coded Partially. Public interest is now assessed in all 12 countries, with 10 Yes and 2 Partially. The distinction between source removal and search-engine de-indexing is clearly explained in eight of 10 assessed countries and partially explained in two.

These findings should not be read as a ranking. A country with more accessible or detailed guidance is not necessarily more protective of privacy, more favourable to delisting requests or more likely to produce a successful outcome. Documentation quality and legal outcome are different questions.

A second important result is that de-indexing and source removal are not the same action. France, Spain, Belgium, Ireland, Austria, Sweden, Poland and other reviewed authorities explain in varying levels of detail that removing a result from a name search does not necessarily remove the underlying article. A publisher controls the original source; a search engine separately controls the search-result processing.

A third result is that the official material consistently treats RTBF as a balancing exercise rather than an automatic deletion right. Public-interest and freedom-of-expression considerations recur across the reviewed guidance. Age matters in several jurisdictions, but age alone is not presented as sufficient. Accuracy matters, but disagreement with an opinion is not automatically the same as factual inaccuracy. Public role and the nature of the underlying event can also affect the analysis.

This report uses a small set of legal and procedural terms repeatedly:

These definitions provide EU-level legal context only. The country coding below remains based on the official national authority sources and scope rules described in the methodology.

The study asked one narrow question: how clearly do national data protection authorities explain the considerations and procedures relevant to Right to Be Forgotten requests involving news articles and search results? It did not test named individuals, submit live RTBF requests, measure response times in practice or assess the likelihood of removal.

The 12 countries were fixed before evidence review. Each country was assessed against eight predefined documentation criteria: journalism/freedom of expression, public interest, age/outdatedness, accuracy/inaccuracy, public role/public figure status, seriousness or nature of the underlying event, source removal versus search de-indexing, and complaint/escalation route.

Evidence was limited to official national data protection authority (DPA) material, including guidance pages, FAQs, procedures, complaint pages and directly relevant DPA decisions or case studies. Search snippets, law-firm commentary, search-engine optimization (SEO) content, Wikipedia and general media coverage were not used as evidence.

For each criterion, the assessed denominator counts Yes, Partially and No clear guidance found values. Unknown / inaccessible fields are excluded and shown separately. No field was coded No clear guidance found because retrieval limitations made it inappropriate to turn missing evidence into an absence claim.

Y = Yes; P = Partially; U = Unknown / inaccessible. These are documentation codes, not outcome scores.

Criteria: 1 journalism/freedom of expression; 2 public interest; 3 age/outdatedness; 4 accuracy/inaccuracy; 5 public role/public figure status; 6 seriousness/nature of underlying event; 7 source removal vs search de-indexing; 8 complaint/escalation route.

Journalism and freedom of expression

Article 17 is not an absolute right. Several authorities expressly connect erasure or delisting with freedom of expression and information, while others apply the same balance through decisions or search-engine guidance. The result is not a simple rule that old or damaging news should disappear. The public's continuing interest in access to the information can remain relevant.

Public interest

Public interest is one of the most consistently documented substantive considerations in the reviewed material. Authorities describe the relevance of the information to internet users, the context of the publication and, in some cases, whether the person remains active in the same professional or public sphere. In Poland, media-specific material from the data protection authority (UODO) discusses balancing media information interests against privacy; because it does not provide a full request-specific delisting test, Poland is coded Partially for this criterion.

Age, accuracy and later developments

Age is frequently relevant, but it is not equivalent to irrelevance. Some official material considers whether information has become outdated, whether later developments changed the context, or whether the information remains relevant to the person's current professional activity. In Austria, the data protection authority (DSB) published an explanation discussing outdated procedural information and context in a delisting judgment. Accuracy is likewise treated carefully: demonstrably inaccurate information can strengthen a request, while disagreement with commentary or opinion may not.

Public role and the underlying event

Several authorities consider whether the individual plays a role in public life and whether the information relates to that role. The nature of the underlying event can also matter. Official examples include distinctions involving professional conduct, political activity, disciplinary matters and criminal proceedings. Austria is coded Partially for the nature of the underlying event, and Poland is coded Partially for public role, because the recovered official material addresses those factors only within a narrower scope. These factors are not universal outcome rules.

Source removal and search de-indexing

This is the distinction most likely to be misunderstood by users. A publisher or website controls the original article. A search engine separately processes links in its search results. A successful name-based delisting request can therefore reduce visibility in search while leaving the source page online and reachable directly, through another query or through another search engine.

For someone dealing with an old news result, the practical sequence is usually more useful than the legal label. The first question is what outcome is actually wanted. If the goal is to stop a result appearing for a name search, the relevant first request is generally directed to the search engine. If the goal is to change, correct or remove the original article, the publisher or source controller is a separate route.

Before submitting a request, preserve the exact result URLs, the source-article URLs, screenshots and the search query used. Where relevant, document later developments, factual inaccuracies, changes in professional role or other context that explains why the result should be reassessed. Keep the original request and every response. Those records are commonly needed if the matter later goes to a DPA.

Under the GDPR, a controller generally has one month to tell an individual what action it has taken on a request to exercise rights under Articles 15-22, subject to a possible extension of up to two further months where the GDPR conditions are met. A DPA complaint is a separate process and should not be described as having the same one-month decision period unless the authority itself says so.

The country summaries below are procedural guides derived from official authority pages and, where noted, limited historical material. They are not predictions of outcome. The listed forms, complaint routes and preparation steps are public procedures that readers can use directly; using Info Remover is not required.

EU legal context

• General Data Protection Regulation (GDPR), Article 4 - definitions including "controller": https://eur-lex.europa.eu/eli/reg/2016/679/art_4/oj
• General Data Protection Regulation (GDPR), Article 17 - Right to erasure ("right to be forgotten"): https://eur-lex.europa.eu/eli/reg/2016/679/

The country list and eight criteria were fixed before final coding. Discovery combined official-domain searches, DPA navigation, rights pages, FAQs, complaint procedures, official PDFs and directly relevant decisions or case studies. Evidence was accepted only where the authority itself explained the relevant point. Party submissions quoted in decisions were not automatically treated as authority reasoning.

Each country-criterion pair was coded Yes, Partially, No clear guidance found or Unknown / inaccessible. Yes required direct and sufficiently clear official support. Partially was used where the authority addressed the point only incompletely, indirectly or in a narrow example. Unknown / inaccessible was used where official evidence could not be reliably recovered. Unknown was never treated as No.

Every newly used source supporting a Yes or Partially value was reopened and checked in a second pass. The final data table, evidence table and country matrix were cross-checked for consistency. This was an AI-assisted research and verification process, not independent human legal review.

Germany required a methodological exception because data-protection supervision is decentralised. The study therefore uses relevant German state-level (Land) and joint supervisory material with explicit attribution rather than treating the federal BfDI as the sole national authority for private-sector search-engine complaints.

The Netherlands also requires a scope note. Five balancing criteria are supported by an official May 2017 AP search-delisting mediation report issued under the former Dutch Data Protection Act (Wet bescherming persoonsgegevens, Wbp), before the GDPR became applicable. Those findings are evidence that the Dutch authority documented those factors historically; they are not presented as a complete current Dutch Article 17 test. The current AP complaint route is separately verified from the present AP website.

After the main coding and pathway audit, a final targeted recovery pass re-examined only the 18 then-unresolved fields. Six were recovered: Austria age/outdatedness and accuracy as Yes, Austria event nature/seriousness as Partially, and Poland public interest, age/outdatedness and public role as Partially. Twelve remained Unknown. The Austrian evidence is a DSB-published explanation of Court of Justice of the European Union (CJEU) judgment C-136/17 rather than a new Austrian DPA decision on an individual complaint. The Polish evidence is the President of UODO's media-conference conclusions and recommendations, so the three recovered fields remain Partially rather than Yes.

The study covers 12 selected EU countries and is not representative of the entire EU. 12 of 96 fields remain Unknown / inaccessible. Missingness is now concentrated in the Netherlands (2), Austria (1), Croatia (4), Malta (3) and Poland (2). This remaining uncertainty does not invalidate the assessed evidence, but it limits broad cross-country claims.

Guidance pages, FAQs, complaint procedures, annual reports and individual decisions are different forms of evidence. A country with a complete or near-complete row may rely on a mixture of these source types. Complete coding therefore does not mean the country has a single comprehensive RTBF guide.

The report measures documentation, not enforcement or legal effectiveness. It does not test complaint outcomes, response quality, actual publisher behaviour, search-engine acceptance rates or the strength of privacy protection. No success rates, country rankings or causal claims should be inferred.

Website content and forms can change. Publication dates and update dates were recorded where visible, but some official pages do not display a clear update date. The report reflects sources checked through 11 September 2026.

The underlying research files are listed alongside the report so readers can see which coding, evidence and procedural records support the narrative.

• country_coding_final_publication.csv - final 12-country coding matrix
• evidence_table_final_publication.csv - field-level evidence, source URLs and scope notes
• recovery_log_final_publication.csv - recovery history
• rtbf_country_pathways_publication.csv - practical country-by-country procedural pathway dataset
• rtbf_pathway_evidence.csv - pathway evidence register
• source_register_final_publication.csv - consolidated official-source register
• remaining_unknowns_final_publication.md - final unresolved fields and scope note

The HTML report at https://info-remover.com/research/eu-right-to-be-forgotten-news-report-2026 is the version of record. The supporting files above are referenced by name; they are not yet published as downloads on this page.

The full claim-level mapping is available in the evidence CSVs. The source register accompanying this report contains all official URLs used for analytical coding and practical pathways. Selected key sources are listed below.