Legal notice

What this archive is

Platform Actions is a structured research archive of enforcement actions taken by digital platforms where AI system involvement is reported or reasonably inferred. It is operated as a public-interest project. Every record cites a publicly identified reporter and one or more independently archived sources. No record is published without source documentation.

The archive documents platform conduct—not the private lives of the people platforms act upon. Where an account identifier necessarily appears in a record, it is incidental to documenting the platform's enforcement decision. The subject of every record is the platform, not the individual.

Legal basis for operation

This archive operates under multiple, independent and overlapping legal bases:

First Amendment (U.S. Constitution). The collection, analysis, and publication of factual information concerning the conduct of powerful private institutions falls within the core of First Amendment protection. Research into AI-assisted enforcement by digital platforms—entities that exercise significant control over public discourse—is a matter of genuine public concern. Judicial enforcement against factual public-interest reporting requires a showing of actual malice that this archive's documented methodology is specifically designed to foreclose.

GDPR—legitimate interests and public-interest archiving. To the extent EU data protection law applies, records are processed on the lawful basis of legitimate interests (GDPR Article 6(1)(f)), given the project's public-interest research and accountability mission. The exemption in Article 17(3)(d) excludes the right to erasure where processing is necessary for archiving in the public interest subject to Article 89(1) safeguards. The dataset constitutes a structured public record of how platforms exercise automated enforcement power over publishers, creators, and individuals. This is a matter of legitimate and documented public interest.

Factual data is not protected expression. Under longstanding U.S. copyright doctrine, facts are not copyrightable. Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991). A platform's enforcement decision — the fact that it occurred, when it occurred, under what stated policy—is not the intellectual property of the platform. Its documentation in a structured database creates no copyright liability.

Copyright and DMCA

This archive does not reproduce copyrighted content. It archives source URLs, not full-text source content. Enforcement screenshots, where held, are enforcement notices issued by the platform to the affected user—notices the user has voluntarily shared as the basis for their report. Their publication in a research context constitutes fair use under 17 U.S.C. § 107: the use is transformative (research and criticism of platform conduct), the amount reproduced is limited to what is necessary to document the action, and the use does not substitute for any market the platform could exploit.

DMCA notice and counter-notice. We comply with the Digital Millennium Copyright Act. Valid notices under 17 U.S.C. § 512(c)(3) will be evaluated on their merits. Counter-notifications are filed for any notice determined to target a legitimate fair-use record. Parties who submit materially false or misleading takedown notices are subject to liability under 17 U.S.C. § 512(f), including damages, attorneys' fees, and costs.

Defamation

Truth is an absolute defense to defamation. Every record in this archive documents what a named person publicly reported, with archived sources. The documented fact is that the report was made and publicly archived—a fact that is true and independently verifiable. Records do not assert that enforcement decisions were correct, incorrect, justified, or motivated by improper conduct; they assert only what happened, as reported.

Platforms are not private individuals. They are powerful institutional actors whose enforcement conduct directly affects public discourse. Under New York Times Co. v. Sullivan, 376 U.S. 254 (1964), a public figure or institution bringing a defamation claim must demonstrate actual malice—knowledge of falsity or reckless disregard for the truth. The methodology of this archive—reported speech, archived sources, explicit epistemic framing, curator review—is specifically designed to document facts accurately, not to assert falsities.

Cross-incident analytical claims appear only in the clearly labeled Findings section, with explicit confidence levels, enumerated supporting records, and documented reasoning. These are statements of documented analytical judgment, not statements of fact about individuals.

Privacy and the right to erasure

GDPR Article 17(1) grants EU/EEA data subjects the right to request erasure of personal data. Article 17(3)(d) exempts processing necessary for archiving in the public interest from this right, subject to appropriate safeguards under Article 89(1).

Records in this archive are retained under the GDPR Article 17(3)(d) public interest exemption (and its equivalent under UK GDPR / Data Protection Act 2018). The incident record, structured fields, and summary are retained even where a verified erasure request is received. EU and EEA data subjects (GDPR Art 17), UK residents (UK GDPR / DPA 2018), and California residents (CCPA) may submit a request through our transparent erasure request process—submit a request here.

What the process involves. Requests may only be submitted by EU and EEA data subjects or California residents who are the subject of the record, or a named, verified legal representative. Identity verification is required: the request must originate from the email address associated with the documented account, or include documented proof of ownership. Platforms, their counsel, and affiliated actors may not submit requests on behalf of subjects. Every request and its outcome is logged in the public erasure request log published with each data export.

Proportionality concession. Where a verified erasure request is received and identity is confirmed, the enforcement screenshot is cleared and the record is annotated to reflect that the subject requested removal. The incident identifier, structured fields, and factual summary are retained under the Article 17(3)(d) exemption—the documentary record of what the platform did is not removed. The identifier will never resolve to a 404; citations to this archive must remain stable by design.

Requests that cannot be verified. Requests that cannot be verified — where identity cannot be confirmed, where the requester is not the subject or a named legal representative, or where the request originates from a platform or affiliated actor—are logged as received, marked as rejected or unverified in the public erasure log, and closed without further action. The subject is notified by email if a valid contact address was provided.

What the project cannot control. The Wayback Machine and archive.today hold their own independent copies of source pages. Historical bulk releases on GitHub Releases are immutable once published (Internet Archive and Hugging Face mirroring are not yet active—see ADR-0005). The erasure-requests.json manifest published with each export is the mechanism for informing downstream data consumers of post-publication requests; it does not reach those releases retroactively.

Data minimisation. We hold only what is necessary to document the enforcement action: platform name, action type, date range, reported account identifier, factual summary, curator commentary, and source URLs. We do not hold private messages, non-public post content, financial information, or health data. Internal curator notes (raw_report) exist for each record—covering chain-of-custody documentation, source authentication assessments, and curation context. These are never published, never included in bulk exports, and never surfaced in the public UI. They are searchable via the project's internal research interface only.

California Consumer Privacy Act. California residents have deletion rights under the CCPA. The project relies on the Section 1798.105(d)(7) exemption (information necessary to comply with a legal obligation) and Section 1798.105(d)(8) exemption (research conducted in the public interest compatible with the purpose for which the personal information was collected). California residents may submit verified deletion requests under the same process as EU/EEA erasure requests—identity verification is required, the enforcement screenshot is cleared on a confirmed request, and the structured record is retained.

Platform complaints

Platforms whose enforcement conduct is documented in this archive have no legal basis to demand its removal. A platform's enforcement decisions—taken at scale, affecting livelihoods and expression—are institutional conduct, not private information. Their documentation in a structured research archive is not a privacy violation, a copyright infringement, or a defamation claim that can survive scrutiny.

Platforms, their counsel, and affiliated actors may not file requests on behalf of subjects. Only the directly affected individual qualifies to raise a factual dispute. Attempts by platforms or their legal representatives to suppress records about their own enforcement conduct are themselves a matter of public interest and are noted in the public record.

Factual disputes

If you have specific, substantiated evidence that a published record contains a factual error—not that you disagree with the platform's original enforcement decision, not that you find the record uncomfortable, but an actual factual error—contact curator@platformactions.org with your name, the record identifier, and the specific claim you dispute along with supporting evidence.

We investigate substantiated factual disputes and correct records where the error is confirmed. Note that the following are not factual errors in a record:

• The platform's enforcement decision itself
• The reporter's account of events
• Curator commentary characterising a pattern
• The existence of a record about you or a platform you represent

Correspondence consisting solely of legal threats without a substantive factual dispute is closed without further response.

Anti-SLAPP notice

Legal claims brought primarily to suppress public-interest research—rather than to redress a genuine legal injury—are Strategic Lawsuits Against Public Participation (SLAPPs). This archive is within the class of expression protected by anti-SLAPP statutes in multiple jurisdictions, including California Code of Civil Procedure § 425.16 and the D.C. Anti-SLAPP Act (D.C. Code § 16-5501 et seq.). These protections apply—including the potential for fee-shifting to the opposing party—in any jurisdiction where they are available.

Governing law

This project is operated from the United States. U.S. federal law and applicable state law govern its operation. To the extent international law imposes additional obligations—including GDPR—we claim all available exemptions for public-interest research and archiving. We will not comply with removal demands based on foreign judgments except where required by an order of a U.S. court with jurisdiction.

Legal process and subpoenas

Legal process directed at this project—subpoenas, court orders, preservation demands—should be addressed in writing to curator@platformactions.org. We will evaluate all legal process for scope, validity, and legal sufficiency, and we will challenge overbroad, unduly burdensome, or constitutionally deficient demands. We will not voluntarily produce information that is not legally required.