This library collects the foundational research, technical specifications, institutional analyses, and regulatory frameworks relevant to the Authenticity Crisis. It is organized by domain and maintained as a permanent reference for researchers, policymakers, journalists, and technologists working on problems of synthetic media, identity verification, content provenance, and the erosion of automatic trust. Entries are selected for structural relevance to the phenomenon, not comprehensiveness. For real-world examples of these dynamics, see the incident archive. For analysis of the research and standards collected here, see the Authenticity Crisis report and the essay collection, including research on synthetic media detection and identity verification infrastructure.
Access labels: Open Access indicates the full text is freely available. Restricted indicates the full text may require institutional access or purchase. PDF indicates a direct link to a downloadable document.
Research Papers
-
GAN-Generated Faces Detection: A Survey and New Perspectives
Survey of detection methods for GAN-generated face images, including deep learning classifiers, physiological signal analysis, and artifact-based approaches, addressing challenges in distinguishing synthetic faces from real human identities.
-
Deepfakes and Cheap Fakes: The Manipulation of Audio and Visual Evidence
Foundational analysis distinguishing between high-fidelity deepfakes and lower-quality manipulations. Establishes the framework for understanding how media manipulation operates across a spectrum of technical sophistication and how even crude alterations erode institutional trust.
-
Adversarial Perturbations Fool Deepfake Detectors
Study demonstrating that adversarial perturbations can significantly reduce the accuracy of deepfake detection systems, highlighting vulnerabilities in current AI-based authenticity verification methods.
-
A Study of the Human Perception of Synthetic Faces
Experimental study examining how humans perceive GAN-generated faces, showing that observers often cannot reliably distinguish synthetic faces from real ones under multiple conditions.
-
Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security
Legal and policy analysis examining the implications of deepfake technology for privacy, democratic institutions, and national security, including risks related to misinformation, reputational harm, and institutional trust.
-
Generative AI Misuse: A Taxonomy of Tactics and Insights from Real-World Data
Systematic classification of generative AI misuse drawn from documented cases. Categorizes tactics including impersonation, fabrication, and manipulation across media types. Provides an empirical basis for understanding the operational patterns that constitute the Authenticity Crisis in practice.
Institutional Reports
-
Increasing Threat of DeepFake Identities
DHS assessment of the operational threat posed by deepfake technology to identity verification systems, border security, and law enforcement. Documents the gap between existing identity infrastructure and the capabilities of generative AI to produce convincing synthetic credentials.
-
Facing reality? Law enforcement and the challenge of deepfakes
European law enforcement analysis of the deepfake threat landscape covering criminal exploitation, detection challenges, and cross-border investigation difficulties.
-
Global Risk Report: AI and Disinformation
Ranks AI-generated misinformation and disinformation as the most severe global risk in the near term. Positions the Authenticity Crisis within the broader context of geopolitical instability, institutional trust erosion, and societal polarization.
-
Synthetic Identity Fraud in the US Payment Ecosystem
Federal Reserve analysis of synthetic identity fraud as a distinct and rapidly growing category of financial crime. Documents how AI-generated identity artifacts defeat existing verification systems, with estimated losses in the billions annually. Directly relevant to the financial dimension of the Authenticity Crisis documented in the documented incident archive.
-
AI-Enabled Influence Operations: Evolving Threat Landscape
Documents the use of generative AI systems in coordinated influence operations, including synthetic text generation for social media manipulation, AI-generated personas, and automated narrative amplification. Provides operational case studies of the Authenticity Crisis deployed as a strategic tool.
Technical Standards
-
Open technical standard for embedding cryptographic provenance metadata in digital media. Defines how images, video, and audio can carry verifiable records of their origin, editing history, and the tools used to create them. Represents the most significant industry effort to build the verification infrastructure the Authenticity Crisis demands.
-
NIST AI Risk Management Framework
Federal framework for identifying, assessing, and mitigating risks associated with AI systems. Includes provisions for synthetic content generation, model transparency, and the societal risks of AI-enabled deception. Provides the institutional vocabulary for addressing the Authenticity Crisis within regulatory contexts.
-
ISO/IEC DIS 27090: Cybersecurity - Artificial Intelligence
International standard addressing cybersecurity risks specific to AI systems, including adversarial attacks on detection systems, model poisoning, and the use of generative models for social engineering. Establishes baseline security requirements for organizations deploying or defending against AI-generated content.
-
Google - written evidence (AIC0012)
Official written evidence submitted to the UK Parliament examining the impact of generative AI on copyright, content authenticity, and information ecosystems. Documents institutional recognition of AI-generated content as a structural challenge requiring legal and technical disclosure frameworks.
W3C Verifiable Credentials Data Model 2.0
Official W3C web standard defining a cryptographically secure and machine-verifiable model for digital credentials. Enables issuers, holders, and verifiers to exchange identity and provenance claims that can be independently verified. Recommended by W3C for wide deployment as core trust infrastructure for the Web.
Identity and Verification Infrastructure
-
The Future of Digital Identity
Official UK government cybersecurity analysis identifying digital identity as foundational infrastructure for secure online services, financial systems, and national digital transformation. Highlights the growing need for robust identity verification mechanisms as AI-generated content and automated systems increase identity spoofing and authentication risks.
-
NIST Digital Identity Guidelines (SP 800-63-4)
Official US federal standard defining technical and procedural requirements for digital identity proofing, authentication, and federation. Updated to address emerging threats including synthetic identity fraud and deepfake-based impersonation. Establishes global reference assurance levels for secure identity verification systems.
-
eIDAS 2.0 European Digital Identity Regulation
Binding European Union regulatory framework establishing interoperable digital identity systems and the European Digital Identity Wallet. Enables citizens to present cryptographically verifiable credentials across member states and provides the legal foundation for trusted identity infrastructure in the age of AI-generated impersonation and synthetic identities.
-
Generative AI and Deepfake Detection in Biometric Systems
Peer-reviewed scientific research examining how generative AI models produce synthetic identities capable of bypassing biometric authentication systems. Documents the emerging security threat posed by deepfakes and highlights the need for new identity verification architectures resistant to AI-generated impersonation.
W3C Decentralized Identifiers (DIDs) v1.0
Official W3C Recommendation defining decentralized identifiers as a new class of globally unique identifiers that do not require a centralized registration authority. Provides the technical foundation for cryptographically verifiable digital identity, enabling individuals and organizations to prove identity and authenticity independently of platforms or governments.
Synthetic Media and Deepfake Analysis
-
FaceForensics++: Learning to Detect Manipulated Facial Images
Benchmark dataset and detection framework for manipulated facial imagery. Established the standard evaluation methodology used by the research community to assess deepfake detection systems and synthetic facial identity manipulation.
-
A Survey of Threats Against Voice Authentication and Anti-Spoofing Systems
Comprehensive survey of modern attacks against voice authentication systems, including deepfake voice cloning, adversarial spoofing, and data poisoning. Documents how synthetic voices generated from minimal audio samples can bypass commercial speaker verification, highlighting systemic weaknesses in biometric identity verification.
-
Peer-reviewed survey analyzing recent advances in AI-generated text detection, proposing a taxonomy of detection approaches and identifying fundamental limitations across current methods. Highlights persistent reliability gaps and the ongoing arms race between increasingly capable generative models and detection systems, reinforcing the structural difficulty of verifying textual authenticity.
-
Cheap Versus Deep Manipulation: The Effects of Cheapfakes Versus Deepfakes in a Political Setting
Peer-reviewed experimental study demonstrating that even low-sophistication synthetic media ("cheapfakes") can influence perception and credibility. Establishes that the authenticity crisis is not limited to advanced deepfakes, but includes a broader spectrum of manipulative synthetic content capable of affecting public opinion and trust.
-
Exploring deepfake technology: creation, consequences and countermeasures
Comprehensive review of deepfake generation, detection, and societal impact. Concludes that increasingly realistic AI-generated media are being misused for deception, coercion, and disinformation, contributing to erosion of trust in digital content and requiring new legal, technical, and institutional countermeasures.
Policy and Regulatory Frameworks
-
The first federal United States law to criminalize the publication of non-consensual intimate imagery generated by artificial intelligence, with penalties of up to two years imprisonment for offenses involving adults and three years for minors. Requires all covered platforms to establish a formal notice and removal process by May 19, 2026, with enforcement by the Federal Trade Commission. Excluded from the definition are broadband internet access providers, email services, and platforms whose content is primarily preselected rather than user-generated. The law's scope is deliberately narrow, addressing nonconsensual intimate imagery while leaving deepfake fraud, political manipulation, and identity fabrication outside its reach. Represents a structural milestone in the legislative recognition of synthetic media as a category of legally actionable harm. Regulatory context is tracked in Signal.
-
AI Legislation Tracker: Deepfakes - US State-Level Bills
Continuously updated tracker of all US state-level legislation targeting AI-generated synthetic media, including bills addressing nonconsensual intimate imagery, electoral deepfakes, and synthetic identity fraud. By early 2026, most US states had enacted deepfake-specific legislation, producing a fragmented regulatory landscape with varying definitions, obligations, and penalties across jurisdictions. Provides essential reference for understanding the scope and limits of the state-by-state approach in the absence of comprehensive federal regulation. The regulatory fragmentation and its relationship to institutional lag are analyzed in Signal.
-
Regulation (EU) 2024/1689 - Artificial Intelligence Act
The first comprehensive legal framework governing artificial intelligence. Establishes binding transparency requirements for AI-generated content, including disclosure obligations for synthetic media and deepfakes, risk classification for high-risk systems, and enforcement mechanisms across EU member states. Represents the first large-scale governmental response to systemic authenticity and identity risks introduced by generative AI.
-
Executive Order 14110: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence
Presidential executive order establishing federal policy and technical standards for AI safety, security, and transparency. Directs NIST and federal agencies to develop watermarking, content provenance, and synthetic media detection standards, recognizing AI-generated content as a national security, fraud, and disinformation risk.
-
No AI FRAUD Act (No Artificial Intelligence Fake Replicas And Unauthorized Duplications Act)
Proposed federal legislation establishing property rights over an individual's voice and likeness, explicitly covering AI-generated voice replicas and digital depictions. Creates civil liability for unauthorized cloning services and synthetic identity impersonation, with statutory damages up to $50,000 per violation. Demonstrates formal legislative recognition of synthetic identity replication as a distinct legal and societal threat.
-
Global Partnership on Artificial Intelligence (GPAI)
Intergovernmental initiative bringing together governments and experts to develop governance frameworks, technical standards, and policy responses to risks posed by artificial intelligence, including synthetic media, identity impersonation, and trust in digital information ecosystems.
-
Online Safety Act: Provisions on Synthetic Content
UK legislation imposing duties on platforms to address harms from synthetic content, including deepfake intimate imagery and AI-generated disinformation. Represents the regulatory approach of embedding synthetic media provisions within broader online safety frameworks rather than creating standalone deepfake legislation.
-
Safeguarding Elections in the Age of AI and Synthetic Content
Official written evidence submitted to the UK Parliament identifying synthetic media and deepfakes as emerging threats to electoral integrity, national security, and democratic trust. Highlights the increasing accessibility, realism, and weaponization of AI-generated content for disinformation, impersonation, and information warfare.
This library is maintained as a living reference, as described on the About page. To suggest additions or report outdated entries, contact signal@authenticitycrisis.com. For ongoing analysis and updates, follow Signal.