# imadethisup.org > A public-education clearinghouse on synthetic media (deepfakes, voice clones, generative imagery) and the law, evidence, and security practice around it, published by Global Cyber Institute, Inc., a U.S. 501(c)(3) nonprofit (EIN 84-2148770). Every claim is linked to a primary source: a peer-reviewed paper, an official statute, a federal agency document, or the originating news investigation. Content is CC BY-NC 4.0; site code is MIT. Single contact: info@imadethisup.org. The site has 70 routes: five core topic pages, a searchable bibliography, an FAQ, a glossary, an interactive tools page, a blog of 31 sourced posts, 20 case summaries each on its own page, and standard organization pages. Each blog post opens with a self-contained "In short" summary and closes with three common questions, both mirrored into schema.org FAQPage markup. There is no advertising, no third-party tracking, and no behavioural profiling. AI crawlers are explicitly permitted in robots.txt. AI use on this site is disclosed in full at /ai-disclosure: content is written with AI assistance under editorial responsibility. Weekly blog posts are researched, drafted and published by an automated agent without a human reading them first, and every citation in those posts is machine-verified against its primary source before publication. Pillar pages, the glossary, references and organisation pages are hand-authored and human-reviewed. No photorealistic imagery on the site is AI-generated. ## Core topic pages - [The Research Lab, how synthetic media is made and detected](https://imadethisup.org/research-lab): Synthetic media is produced by generative architectures, principally generative adversarial networks and latent diffusion models, and each architecture leaves characteristic artifacts that forensic detection methods exploit. Detection remains unsolved because a detector trained on one generator frequently fails on output from another, a limitation known as the cross-generator generalization problem. - [The War Room, operational defense against deepfake fraud](https://imadethisup.org/war-room): Synthetic voice and video are now inexpensive enough for ordinary fraud crews, which makes business email compromise and executive impersonation a process problem rather than a detection problem. The control that defeats them is out-of-band callback verification on payment instructions above a set threshold, applied regardless of how convincing the request sounds. - [The Safety Suite, personal defense and takedown routes](https://imadethisup.org/safety-suite): Most synthetic-media harm is interpersonal: harassment, image-based abuse, and fraud against family members. United States law changed substantially from 2024 onward, the TAKE IT DOWN Act creates a federal prohibition on non-consensual intimate imagery together with a 48-hour platform takedown obligation, and hash-matching services can block re-uploads across participating platforms. - [Provenance, C2PA, Content Credentials, and signing the truth](https://imadethisup.org/provenance): Provenance establishes where a media asset came from and what was done to it, using cryptographic signatures applied at creation rather than analysis applied afterwards. C2PA Content Credentials is the leading open standard and JPEG Trust is the ISO complement. Provenance does not prove an image is true, it proves what a signer asserted. - [References & Standards, the full sourced bibliography](https://imadethisup.org/references): The imadethisup.org reference collection holds every source cited across the site: peer-reviewed papers, official technical standards, federal agency documents, and originating news investigations. Each entry links to a primary source rather than to secondary coverage, and entries can be filtered by audience or searched by keyword. Broken links can be reported to the editorial mailbox. - [Case law, what courts have held about deepfakes and fabricated evidence](https://imadethisup.org/case-law): United States courts have begun deciding cases about synthetic media directly. Decisions since 2024 address whether video evidence can still be authenticated by circumstantial detail in an era of deepfakes, when AI-generated intimate imagery is criminal, what sanctions follow filing citations a language model invented, and whether an AI voice clone of a candidate is actionable under telephone and election law. ## Resource pages - [Frequently Asked Questions](https://imadethisup.org/faq): The imadethisup.org FAQ answers common questions about synthetic media, deepfakes, generative-AI fraud, detection methods, content provenance, and United States law. Answers are grouped by subject, and each carries an inline citation to a primary source with a link to the longer treatment elsewhere on the site where a short answer is not sufficient. - [Glossary of synthetic-media and forensics terms](https://imadethisup.org/glossary): The imadethisup.org glossary defines the synthetic-media, deepfake, generative-AI, and digital-forensics terms used across the site, in the specific sense the site uses them. Where a term originates in a paper, a technical standard, or a statute, the primary source is cited inline. Entries are searchable and range from business email compromise to C2PA. - [Tools, interactive demos and trusted third-party services](https://imadethisup.org/tools): imadethisup.org provides free interactive tools for synthetic-media literacy with no signup required: a deepfake calibration quiz that measures a reader's own detection accuracy, and a real-versus-synthetic comparison slider. The tools page also lists trusted third-party services for verifying content provenance and for removing non-consensual intimate imagery, including StopNCII and NCMEC. - [Blog, sourced case analyses, research summaries, policy explainers](https://imadethisup.org/blog): The imadethisup.org blog publishes short, sourced notes on synthetic media: case analyses, research summaries, and policy explainers. Every post cites primary sources, statutes, court filings, peer-reviewed papers, and federal agency documents, and links to the same reference collection used across the rest of the site. The full archive is available as an RSS feed. ## Engage - [Retain a deepfake expert](https://imadethisup.org/retain-an-expert): Global Cyber Institute coordinates introductions to independent experts on deepfakes and synthetic media, covering litigation and expert-witness work, internal investigations, corporate consulting, and press commentary. Inquiries submitted through the form on this page are routed to a vetted examiner. The organisation is a United States 501(c)(3) nonprofit and the experts are independent. ## Organization - [About this project](https://imadethisup.org/about): imadethisup.org is published by Global Cyber Institute, Inc., a United States 501(c)(3) nonprofit (EIN 84-2148770) advancing research and literacy in cybersecurity, digital forensics, and synthetic-media risk. The site is edited anonymously as a matter of policy, carries no advertising or sponsorship, and uses no third-party tracking. Editorial content is licensed CC BY-NC 4.0. - [AI Disclosure, how AI is used in producing this site](https://imadethisup.org/ai-disclosure): imadethisup.org is written with AI assistance and published under editorial responsibility. Large language models help with research triage, drafting, and structural editing. Weekly blog posts are researched, drafted and published by an automated agent without a human reading them first; every citation in those posts is machine-verified against its primary source, and a post whose sourcing fails that check is not published. Pillar pages, the glossary, references and organisation pages are hand-authored and human-reviewed. No photorealistic imagery on this site is AI-generated. - [Contact](https://imadethisup.org/contact): For press inquiries, corrections, collaboration, takedown questions, and privacy requests, write the same address. Global Cyber Institute, Inc. does not maintain a public mailing address; correspondence is conducted by email. - [Disclaimer](https://imadethisup.org/disclaimer): Educational content only. Nothing here is legal, investigative, technical, or professional advice, and reading it forms no relationship. - [Privacy Policy](https://imadethisup.org/privacy): Minimum data collection, no third-party trackers, no behavioural advertising. Data-rights instructions for the EU, UK, and several U.S. states. - [Privacy Choices, exercising GDPR / CCPA rights](https://imadethisup.org/privacy-choices): A practical guide to exercising your data-protection rights on this Site, under GDPR (EU/UK), CCPA/CPRA (California), and the state privacy laws of Virginia, Colorado, Connecticut, Utah, and Texas. - [Sitemap](https://imadethisup.org/sitemap): Machine-readable sitemap is at /sitemap.xml. ## Blog posts All 31 posts, newest first. Each is independently sourced with inline citations. - [Reporting non-consensual imagery: what one audit found](https://imadethisup.org/blog/reporting-ncii-to-platforms-what-research-shows), 2026-08-30: A 2024 audit study compared two reporting mechanisms available to victims of non-consensual intimate imagery: reports filed under non-consensual nudity violations and reports filed as copyright infringement [1]. The study uploaded AI-generated nude images to X, formerly Twitter, and tracked removal speed across both routes. The copyright infringement mechanism worked faster than the platform's dedicated non-consensual nudity mechanism [1]. Because the study tested AI-generated content on one platform, it measures reporting flow performance rather than addressing real victims' images or other platforms. - [What invisible watermarks can and cannot survive](https://imadethisup.org/blog/how-robust-is-invisible-watermarking), 2026-08-30: Invisible watermarks on AI-generated images face three distinct threats: removal (deleting an existing mark), forgery (adding a mark to unmarked content), and detection degradation (making marks harder to verify through image processing). Removal attacks can work across multiple generative models without customisation. Forgery requires only a single watermarked example and no knowledge of the watermarking scheme. Detection can degrade under standard image distortions and adversarial attacks. Watermarks raise the cost of misrepresentation but do not guarantee detection. Certified watermarking methods with provable robustness bounds exist, shifting the question from whether watermarking works to under what conditions it works. - [Detector results on deepfakes found in the wild](https://imadethisup.org/blog/deepfake-detectors-on-real-world-deepfakes), 2026-08-29: Deepfake-Eval-2024 is a benchmark assembled from deepfakes that actually circulated online during 2024, rather than generated for evaluation. Open-source detectors that score highly on academic datasets drop sharply on it, with reported AUC decreasing by 50 percent for video, 48 percent for audio and 45 percent for image models. Commercial and finetuned models do better without closing the gap to human analysts. - [Penalties for breaching the EU AI Act's disclosure rules](https://imadethisup.org/blog/eu-ai-act-article-50-penalties), 2026-08-29: Breaches of the EU AI Act's transparency obligations for synthetic content are penalised under Article 99(4)(g), which sets administrative fines of up to EUR 15,000,000 or 3 percent of total worldwide annual turnover, whichever is higher. Enforcement falls to Member State authorities rather than to the European Commission. For SMEs and start-ups the cap is whichever of those two figures is lower. - [Synthetic media disclosure obligations now in force](https://imadethisup.org/blog/eu-ai-act-article-50-deepfake-disclosure), 2026-08-29: Article 50 of the EU AI Act requires providers of AI systems to mark synthetic audio, image, video, and text in machine-readable format. Deployers must disclose deepfakes and AI-generated text about public-interest matters. A key exemption applies: public-interest text subject to human review or editorial control, with a named person or legal person holding editorial responsibility, need not be disclosed. - [What a deepfake detector score does not tell you](https://imadethisup.org/blog/why-detector-scores-do-not-travel), 2026-08-29: Deepfake detector scores describe performance on a specific benchmark, and published figures fall substantially when the same detectors meet generators they were not trained on. One chronological evaluation reports that generalisation to future generators, without retraining, is near-random. Anyone weighing a detector output for a report, a filing or a story needs to know which generators it was trained on. - [What Content Credentials establish, and what they do not](https://imadethisup.org/blog/what-content-credentials-establish), 2026-08-29: Content Credentials are tamper-evident, cryptographically signed records of how a file was created and edited, travelling with the asset. Reading one tells you what a signer asserted about the file's history. Content Credentials do not establish that the depicted events occurred, do not identify the person who used a tool, and their absence from a file is uninformative, because ordinary editing and re-encoding routinely strip metadata. - [What the C2PA conformance registry actually lists](https://imadethisup.org/blog/which-cameras-have-certified-content-credentials), 2026-08-29: The C2PA Conforming Products List is a public registry recording which products have passed the C2PA conformance program, at what assurance level, and against which version of the specification. Reading it directly is the difference between accepting a provenance claim and checking one. As retrieved on 29 August 2026 it held 174 conformant records. - [What the US AI Safety Institute says still needs research](https://imadethisup.org/blog/synthetic-content-research-gaps), 2026-08-29: The US AI Safety Institute identifies four technical research gaps: digital watermarking robustness, content authentication mechanisms, detection of synthetic materials, and safeguards preventing harmful model outputs. It also calls for social science research on how synthetic content affects information integrity, public trust, sensitive domains like education and counselling, and strategies against fraud and impersonation. - [When audio deepfake detectors fail on new voice generators](https://imadethisup.org/blog/audio-deepfake-detection-generalization), 2026-08-29: Synthetic speech detectors encounter a fundamental generalization problem: systems trained on one set of voice generators degrade significantly when tested on generators they were not designed to detect. The gap is not simply that unseen generators produce harder examples to classify; they produce fundamentally different acoustic patterns. This means that boosting model capacity alone will not solve the problem. Detection results should be interpreted cautiously if the report does not confirm that the detector was trained on, or tested against, the specific synthesis method in question. - [Why deepfake impersonation works, and what stops it](https://imadethisup.org/blog/why-deepfake-impersonation-defeats-detection), 2026-08-29: Deepfake impersonation of executives works by manipulating trust cues and decision-making processes, not by deceiving the eye alone. The durable control is an out-of-band verification step, such as a separate phone call or in-person confirmation, that does not depend on detecting whether the media is real. - ["Follow the money" is no longer enough](https://imadethisup.org/blog/follow-the-money-ai-fraud), 2026-06-15: Transaction tracing still matters, but it no longer proves the case on its own. When the inducement is a cloned voice and the laundering runs through cross-chain bridges, tracing can establish flow and identify beneficiaries, yet it cannot show what people knew, what they said to each other, or how decisions were made. - [Agentic AI and accountability: who answers when an agent trades?](https://imadethisup.org/blog/agentic-ai-accountability-sec), 2026-06-15: When an autonomous AI agent violates the securities laws, the firm that deployed it answers. Scienter can still be established by aggregating knowledge across corporate actors and imputing recklessness to a firm that permissions an agent into sensitive systems without controls, and where no individual's intent can be proven, common-law agency principles keep the principal liable. - [AI governance from the security chair: what the CISO owns](https://imadethisup.org/blog/ai-governance-from-the-security-chair), 2026-06-15: AI governance increasingly lands on the security leader's desk. The remit is defined by three texts, the NIST AI Risk Management Framework, its Govern/Map/Measure/Manage functions being the spine, and is executed through three control families: logging, evaluation, and provenance. For public companies, board reporting on this risk is no longer discretionary. - [AI tools and attorney-client privilege: keeping confidences confidential](https://imadethisup.org/blog/ai-tools-attorney-client-privilege), 2026-06-15: Feeding client information to a third-party AI model implicates two distinct protections. ABA Model Rule 1.6 bars revealing information relating to the representation and requires reasonable efforts to prevent disclosure; separately, privilege protects only communications kept confidential, so voluntary disclosure to a third party outside the circle of representation generally waives it. - [Authenticating AI-touched evidence: do we need a new rule?](https://imadethisup.org/blog/authenticating-ai-evidence-rule-901), 2026-06-15: Federal Rule of Evidence 901(b)(9), evidence describing a process or system that produces an accurate result, is the workhorse for machine output, but its bar is low enough that a competent fake can clear it. Reform proposals would require a "valid and reliable" result, and a proposed Rule 707 would apply Rule 702 reliability standards to machine-generated evidence offered without a sponsoring expert. - [Building a deepfake incident-response plan for smaller firms](https://imadethisup.org/blog/deepfake-incident-response-plan), 2026-06-15: A synthetic-media incident often leaves no technical footprint, no breached server, just a call or message that persuaded someone. Anchor the response to NIST SP 800-61 Revision 3 and run five steps: detect, contain, preserve, notify, recover. Containment means calling the bank's fraud line immediately, because wire-recovery windows are measured in hours, not days. - [Deepfake defense for real-estate closings, at the process layer, not the model.](https://imadethisup.org/blog/deepfake-wire-fraud-real-estate), 2026-06-15: Voice and video impersonation now reaches title, escrow, and closing workflows, in September 2024 a Florida title company nearly wired roughly USD 250,000 to a fraudster who joined a Zoom call posing as the property owner. Detection is a useful layer but only a probability; the control that works is refusing to authorize funds without an out-of-band callback on the wiring instructions. - [Deepfakes and the integrity of evidence in family court.](https://imadethisup.org/blog/deepfakes-in-family-court-evidence), 2026-06-15: A fabricated video can do its damage in a custody dispute before anyone proves it fake. The remedy is not better detection but an authentication discipline the rules already support: Federal Rule of Evidence 901(a) has always required a proponent to produce evidence sufficient to support a finding that the item is what they claim it is. - [Deepfakes are reshaping workplace sexual harassment.](https://imadethisup.org/blog/deepfakes-workplace-harassment), 2026-06-15: Synthetic intimate imagery of a coworker turns harassment into a remote, repeatable act, and Title VII of the Civil Rights Act of 1964 does not require the conduct to occur at the office. Harassment becomes the employer's problem when it is severe or pervasive enough to affect the terms and conditions of employment, wherever the image was made. - [Forensic neutrals: who decides when digital evidence is contested?](https://imadethisup.org/blog/forensic-neutrals-digital-evidence), 2026-06-15: When adversaries cannot agree whether a file or an AI-generated artifact is authentic, courts have an underused option: appoint a neutral. Federal practice offers two doors, a court-appointed expert under Federal Rule of Evidence 706, and a special master under Rule 53, with the master's independence enforced by the same disqualification standard that applies to judges. - [From code to canvas: the copyright fight over generative-AI creations.](https://imadethisup.org/blog/generative-ai-copyright-canvas), 2026-06-15: Generative AI raises two separate copyright questions with two different answers. On the output side the law is settled: copyright protects human authorship, so purely machine-generated elements are not protectable and must be disclaimed on registration. On the input side, whether training on copyrighted works is fair use, the law is genuinely unsettled. - [Investigating white-collar fraud when the evidence can be fake](https://imadethisup.org/blog/white-collar-fraud-investigation), 2026-06-15: Complex frauds surface as friction, a variance that will not reconcile, a vendor whose invoices follow a suspicious pattern. In the AI era the documents explaining that friction may themselves be synthetic, so authenticity becomes a separate line of inquiry rather than an assumption, and the dataset an investigator is offered is not the dataset that exists. - [Reg S-P's new baseline for smaller firms](https://imadethisup.org/blog/reg-s-p-small-firm-data-security), 2026-06-15: As of June 3, 2026, smaller SEC-registered advisers and broker-dealers must comply with the amended Regulation S-P. Three obligations apply: a written incident-response program, notification of affected customers as soon as practicable and no later than 30 days, and service-provider oversight requiring vendors to report breaches within 72 hours. Outsourcing the data does not outsource the duty. - [The hallucination tax: what unverified AI citations really cost](https://imadethisup.org/blog/ai-hallucination-tax-arbitration), 2026-06-15: A fabricated AI citation is not a software glitch but a breach of the duty of competence, and it levies a real cost, sanctions, wasted hours, forfeited credibility. ABA Model Rule 1.1 Comment 8 already requires lawyers to understand the risks of relevant technology; no rule change was needed to make verification mandatory. - [Using AI in e-discovery: how to defend the review, not just run it](https://imadethisup.org/blog/ai-ediscovery-in-arbitration), 2026-06-15: Courts settled the threshold question over a decade ago: technology-assisted review is an acceptable way to search for relevant ESI. What gets challenged now is validation, whether the producing party can demonstrate the result was adequate, typically through recall and precision measures, and whether a human attorney can certify the response under FRCP 26(g). - [When the dispute is about AI: arbitration's new rules for AI discovery and evidence](https://imadethisup.org/blog/jams-ai-rules-discovery), 2026-06-15: The JAMS AI Disputes Rules answer a problem litigation rules were not built for: how to take discovery of a model that learns and drifts. The signature move is to take the experts to the model rather than the model to the parties, proprietary hardware, software, models, and training data are made available to experts rather than handed over. - [When the file lies about itself.](https://imadethisup.org/blog/ai-metadata-hallucination), 2026-06-15: Generative tools can write false author, date, and provenance values into the files they produce. Because metadata is treated as evidence in discovery and authenticity disputes, a fabricated Author or Created field can mislead at two depths: the ordinary Properties pane a reviewer reads, and the deeper fields forensic tools parse. Verify provenance; do not trust it. - [What the Arup deepfake actually proves, and what it doesn't.](https://imadethisup.org/blog/arup-deepfake-what-it-proves), 2026-04-27: The Arup Hong Kong fraud succeeded because a video call could substitute for payment authorization, not because the deepfake was undetectable. Arup's CIO stated no systems were compromised and no data was affected; it was "technology-enhanced social engineering." The break point was the payment-approval workflow, and the missing control was an out-of-band callback. - [What the TAKE IT DOWN Act actually changes, and what it doesn't.](https://imadethisup.org/blog/take-it-down-act-explained), 2026-04-27: The TAKE IT DOWN Act (S.146) prohibits knowingly publishing, without consent, intimate visual depictions of minors or non-consenting adults, and deepfakes intended to cause harm. The criminal prohibition took effect on enactment. It gives survivors a federal floor for platform takedown speed, 48 hours, but does not amend Section 230. - [Why deepfake detectors fail on new generators.](https://imadethisup.org/blog/why-detectors-fail-on-new-generators), 2026-04-27: Deepfake detectors fail on unfamiliar generators because they learn an architecture's fingerprint rather than synthesis itself. Frank et al. (ICML 2020) showed GAN-generated images carry severe frequency-domain artifacts caused by upsampling operations, so a detector trained on them is really learning the upsampling signature of one family of decoders, and it does not transfer. ## Case summaries 20 United States decisions on deepfakes, AI-generated evidence and fabricated citations. Each has its own page summarising the decision in IRAC form, written from the opinion itself, and linking to the full text. - [Brasse v. State (Md. App. Ct. 2025) · imadethisup.org](https://imadethisup.org/case-law/brasse-v-state-md-app-2025): Maryland's child pornography statute is not facially overbroad, because section 11-208 reaches only images of an actual child and computer-generated ima... - [Cassata v. Macrina Architect (2026) · imadethisup.org](https://imadethisup.org/case-law/cassata-macrina-architect-suffolk-county-firm-and-supervisor-sanctions): A lawyer who copies another lawyer's brief into her own papers adopts its fabricated citations as her own, and under 22 NYCRR 130-1.1 the supervising at... - [Estate of Caviness v. Atlas Air (2026) · imadethisup.org](https://imadethisup.org/case-law/caviness-akerlund-atlas-air-eleventh-circuit-hallucinated-briefs): A lawyer who signs Eleventh Circuit briefs filled with AI-generated citations to nonexistent cases, and who then purports to withdraw the wrong set of c... - [Day v. Thompson (La. 2025) · imadethisup.org](https://imadethisup.org/case-law/day-v-thompson-surveillance-video-in-camera-review): A Louisiana trial court cannot perform the La. C.E. art. 607(D)(2) balancing test on surveillance video without first reviewing the video in camera, and... - [Dineen/Shibata v. Kotchka (2026) · imadethisup.org](https://imadethisup.org/case-law/dineen-shibata-v-kotchka-self-represented-litigant-arizona-sanction): A self-represented Arizona litigant who files an opening brief containing hallucinated case citations produced by generative AI, and who does nothing to... - [In re Irizarry Centeno (P.R. 2026) · imadethisup.org](https://imadethisup.org/case-law/in-re-irizarry-centeno-bonilla-irizarry-puerto-rico-no-ethical-sanction): Filing court papers containing non-existent or misattributed citations produced with technological assistance is sanctionable conduct in Puerto Rico und... - [Jigsaw Productions v. SEC (D.D.C. 2026) · imadethisup.org](https://imadethisup.org/case-law/jigsaw-productions-v-sec-deepfake-risk-foia-exemption): A federal agency cannot satisfy the Freedom of Information Act's foreseeable harm requirement by asserting that a released recording could be manipulate... - [League of Women Voters v. Kramer (2025) · imadethisup.org](https://imadethisup.org/case-law/league-of-women-voters-v-kramer-ai-robocall-motion-to-dismiss): The District of New Hampshire held that an AI voice-cloned robocall telling voters that casting a primary ballot would waste their general election vote... - [Matter of Gabriel H. (2024) · imadethisup.org](https://imadethisup.org/case-law/matter-of-gabriel-h-dk-fourth-department-deepfake-weight): The Appellate Division, Fourth Department held that three hacked security camera videos were sufficiently authenticated by circumstantial evidence of th... - [Matter of M.S. (M.H.) (N.Y. Ct. App. 2026) · imadethisup.org](https://imadethisup.org/case-law/matter-of-ms-mh-deepfake-video-authentication): The New York Court of Appeals held that Erie County failed to authenticate three videos under People v Patterson where neither the person who allegedly... - [Mooney v. State (Md. 2024) · imadethisup.org](https://imadethisup.org/case-law/mooney-v-state-video-authentication-circumstantial-evidence): The Supreme Court of Maryland held that video footage may be authenticated through circumstantial evidence under Maryland Rule 5-901(b)(4), including a... - [Noland v. Land of the Free (2025) · imadethisup.org](https://imadethisup.org/case-law/noland-land-of-the-free-california-ai-fabricated-quotations): No brief or other paper filed in any California court should contain a citation, whether supplied by generative AI or any other source, that the attorne... - [Pegasystems v. Appian (Va. Ct. App. 2024) · imadethisup.org](https://imadethisup.org/case-law/pegasystems-v-appian-authenticating-software-evidence): Electronic evidence in Virginia is authenticated under the same very modest Rule 2:901 standard as anything else, the possibility that electronic eviden... - [Pennington v. First Hand Land (2026) · imadethisup.org](https://imadethisup.org/case-law/pennington-v-first-hand-land-forged-court-order-sanction): Submitting a blatantly forged court order as a basis for relief is conduct utterly inconsistent with the orderly administration of justice, and the Dist... - [Prososki v. Regan (Neb. 2026) · imadethisup.org](https://imadethisup.org/case-law/prososki-regan-nebraska-supreme-court-fictitious-authority): Submitting fictitious authority to a Nebraska court is resolved under the existing rules of professional conduct and court rules whether or not generati... - [State v. Amyda (Iowa Ct. App. 2026) · imadethisup.org](https://imadethisup.org/case-law/state-v-amyda-deepfake-claim-video-authentication): The Iowa Court of Appeals held that a digital video may be authenticated under Iowa Rule of Evidence 5.901(b)(4) entirely on circumstantial evidence of... - [State v. Currie (2025) · imadethisup.org](https://imadethisup.org/case-law/state-v-currie-deepfake-objection-probation-search-photo): The Ohio First District held that a probation officer relying on a social media photograph to establish reasonable grounds for a probation search is not... - [State v. Dore (Conn. App. Ct. 2026) · imadethisup.org](https://imadethisup.org/case-law/state-v-dore-conn-app-2026): Connecticut's second degree child pornography statute is not overbroad, because General Statutes section 53a-193 (13) limits it to depictions of real ch... - [United States v. Anderegg (7th Cir. 2026) · imadethisup.org](https://imadethisup.org/case-law/united-states-v-anderegg-7th-cir-2026): Because the AI-generated images charged against Anderegg depict no actual child, Stanley v. Georgia and Ashcroft v. Free Speech Coalition bar prosecutin... - [United States v. Schram (8th Cir. 2025) · imadethisup.org](https://imadethisup.org/case-law/united-states-v-schram-8th-cir-2025): A jury may determine from the images themselves that advertised child sexual abuse material depicts real children, and the government need not produce e... ## Machine-readable indexes - [Full text export](https://imadethisup.org/llms-full.txt): the complete prose of every page on the site in one file. - [Sitemap](https://imadethisup.org/sitemap.xml): XML sitemap with last-modified dates. - [RSS feed](https://imadethisup.org/feed.xml): all blog posts. - [Robots policy](https://imadethisup.org/robots.txt): AI crawlers explicitly allowed. ## Citation If you reproduce or summarise this material, the CC BY-NC 4.0 licence requires attribution. Please cite "imadethisup.org, a project of Global Cyber Institute, Inc." and link to the specific page so readers can reach the primary sources themselves.