Art & Culture Law · 2026-04-20

📰 Art & Culture Law — 2026-04-20

📚 Anna's Archive Faces $322M Default Judgment as Spotify and Labels Win Shadow Library Battle
🔧 GitHub's DMCA Removal Record and Cox v. Sony Signal Shifting Copyright Terrain
🏛️ Delhi High Court Fractures India's World-Leading Site-Blocking Architecture
🎮 Playdate Catalog Bans Generative AI as Games Industry Solidifies Anti-AI Consensus
🇪🇺 EU Think Tank Calls for IP-Blocking Ban and Rightsholder Liability Reform
🖼️ 11th Circuit Revives Frida Kahlo Corp. Trademark Fight Against Family Claimants

---

📚 Anna's Archive Faces $322M Default Judgment as Spotify and Labels Win Shadow Library Battle

Anna's Archive, the meta-search engine for shadow libraries that quietly became the world's largest repository of pirated books, crossed a line last December when it announced the backup of Spotify's entire catalog — a 300TB music archive intended for eventual public release. The move triggered an immediate legal response: Spotify, Universal Music, Sony Music, and Warner joined forces in a lawsuit filed days after the announcement, obtaining preliminary injunctions targeting the site's domain registrars. The shadow library lost several domains, added a Greenland .gl backup address, and quietly removed the Spotify music from its torrents page, hoping removals would defuse the case.

They were not enough. On April 15, Judge Jed Rakoff of the Southern District of New York entered a $322 million default judgment against the site's unknown operators — the maximum available under the labels' damages calculation. The structure is revealing: Warner, Sony, and Universal each claimed $150,000 statutory damages per sound recording for willful infringement, totaling $22.2 million across 148 recordings. Spotify added DMCA circumvention damages of $2,500 for 120,000 confirmed-released music files — $300 million of the total. Applied to the full 2.8 million released files, liability would exceed $7 billion.

The judgment is simultaneously a legal victory and an enforcement fiction. Anna's Archive's operators remain unidentified. Rakoff ordered them to file a compliance report within ten business days including contact information, under penalty of perjury — almost certain to go ignored. The permanent worldwide injunction covers ten Anna's Archive domains (.org, .li, .se, .in, .pm, .gl, .ch, .pk, .gd, .vg), directing registrars including Cloudflare, Public Interest Registry, and Njalla SRL to disable access, suspend nameservers, and cease hosting. Domains under registries outside U.S. jurisdiction — the .gl address proved resistant even to preliminary orders — may not comply.

The cultural infrastructure stakes run deeper than the music-industry framing suggests. Anna's Archive positions itself as a preservation project: its Spotify mirror was framed as a hedge against platform collapse. The legal system treats this as irrelevant. Preservation intent creates no copyright exemption, and circumvention of Spotify's DRM generates liability independent of the underlying infringement claim. The gap between the $322M judgment and the $0 collectible from anonymous global operators illuminates the structural problem of copyright enforcement against pseudonymous infrastructure: the law can condemn, but the internet routes around verdicts. What survives the judgment is not a resolved dispute but a legal record of what shadow libraries cost — in theory.

Sources:

---

🔧 GitHub's DMCA Removal Record and Cox v. Sony Signal Shifting Copyright Terrain

GitHub's 2025 DMCA transparency report, released April 19, reveals the platform processed 2,661 takedown notices last year — up roughly 33% from 2024 — affecting 47,228 repositories, a 51.6% increase. The distribution is stark: August and November alone accounted for nearly half the year's total removals (12,030 and 11,357 repositories respectively), strongly suggesting a small number of bulk complaints against heavily forked projects rather than a broad enforcement surge. Anti-circumvention claims — notices targeting projects that bypass copy controls under DMCA Section 1201 — jumped 41% to 645, up from 63 in 2020.

The circumvention number is strategically significant. GitHub's policy requires circumvention claims to be "carefully reviewed by legal experts and engineers," erring on the side of developers when validity is unclear. The $1 million Developer Defense Fund — established after the RIAA's 2020 takedown of youtube-dl — backs developers against overreach. The surge in anti-circumvention claims, concurrent with AI training dataset disputes generating novel circumvention theories, marks an expanding legal frontier.

The report landed as GitHub publicly celebrated the Supreme Court's Cox v. Sony ruling, which it called a "landmark victory for the open internet." The Court reinforced that service providers are not automatically liable for user infringement absent evidence of intent to encourage or materially contribute to it — a significant narrowing of secondary liability theories that rightsholders had leveraged aggressively since Grokster. GitHub's Margaret Tucker wrote that the ruling "protects developers from copyright enforcement overreach" and gives platforms room to side with users in ambiguous cases.

The convergence creates a productive paradox: the DMCA system is simultaneously under record volume pressure and more legally fortified against abuse than at any point since the statute's enactment. The 86% rejection rate that WordPress reported for its own DMCA notices — with AI-generated takedown notices flagged as a new manipulation vector — confirms that enforcement inflation is outpacing enforcement quality. Cox v. Sony gives platforms cover to resist bulk claims without the secondary liability fear that historically forced capitulation. For repositories handling AI training data, model weights, and circumvention-adjacent tooling, this cover is operationally real. The meaningful shift is not the record notice count; it is the changed calculus platforms can now apply when assessing whether a given claim is valid or simply voluminous.

Sources:

---

🏛️ Delhi High Court Fractures India's World-Leading Site-Blocking Architecture

India's "Dynamic+" site-blocking regime — the most aggressive IP enforcement architecture on earth — has hit a constitutional ceiling in its own originating court. The Delhi High Court, which pioneered dynamic blocking injunctions targeting 1337x, The Pirate Bay, and others beginning in 2019 and subsequently extended them to U.S. domain registrars like Namecheap, has now issued a ruling that directly conflicts with its own precedent, creating a judicial fracture requiring a higher bench to resolve.

The case began as a routine trademark dispute. Mahindra and Mahindra sued packers-and-movers businesses using the MAHINDRA mark, obtained blocking orders against five infringing domains, and requested the court make the order "future-proof" by allowing a court official to add newly discovered mirror domains without returning to a judge — exactly the procedure Dynamic+ injunctions have used routinely.

Justice Tushar Rao Gedela said no. The reasoning is procedural but foundational: once a judge signs a final ruling and closes a case, the court's authority over that matter ends. A court officer cannot add new parties or extend injunctions post-closure. The practice of delegating domain-addition authority to an official — standard in Dynamic+ orders — is, in Gedela's framing, "beyond comprehension." Rightsholders must initiate new proceedings or use execution proceedings for new domains; neither provides the real-time blocking speed that made Dynamic+ valuable.

SpicyIP's Tejaswini Kaushal notes the ruling creates a direct divergence between judges of the same court — a rightsholder appearing before a different Delhi HC bench could receive the opposite answer today. Gedela himself called the situation "urgent and alarming" and directed Parliament to update civil procedure rules and intermediary regulations to create a proper statutory basis for post-judgment blocking. Until Parliament acts or a higher bench resolves the conflict, India's system operates in legally uncertain territory.

The structural implications extend globally. Dynamic+ orders had become a model for IP enforcement precisely because they solved the hydra problem: new domains blocked before they gain traffic. The regime also extended to U.S. registrars, making it geopolitically significant for global enforcement strategies. A fractured Dynamic+ architecture in the world's most blocking-intensive jurisdiction signals that even the most aggressive enforcement regimes eventually encounter the procedural limits of adjudicative power. The infrastructure of copyright enforcement — real-time, automatic, extra-jurisdictional — runs ahead of the legal authority that supposedly governs it.

Sources:

---

🎮 Playdate Catalog Bans Generative AI as Games Industry Solidifies Anti-AI Consensus

Panic, creator of the Playdate handheld console, revised its Catalog developer policy this month to prohibit generative AI in game art, audio, music, text, and dialog — the most comprehensive platform-level content ban yet from a mainstream games publisher. The policy went into effect in April 2026 with a narrow carve-out permitting games that used AI in coding, which are disclosed to players. Cofounder Cabel Sasser stated plainly that Panic "does not have any interest in generative AI-created products."

The Playdate ban crystallizes a trend GDC data has been tracking for three years. A recent GDC developer survey found 52% of respondents believe generative AI is having a negative impact on the game industry — up from 30% in 2025 and 18% in 2024. The trajectory is as significant as the absolute number: industry sentiment against generative AI has nearly tripled in two years, even as AI vendor presence at GDC has expanded dramatically. Vendors pitched AI-driven NPCs and AI-generated worlds built from chat prompts; the developers making actual games largely refused.

The aesthetic rejection intersects with legal reality. AI-generated art cannot be copyrighted after the Supreme Court in March 2026 declined to hear Stephen Thaler's appeal — affirming that "human authorship is a bedrock requirement of copyright." Studios selling games commercially need protectable assets. Finji cofounders Adam and Rebekah Saltsman said they don't believe there's a legal framework to actually sell a game built from generative AI output. Legal exposure compounds the aesthetic rejection: the same output that developers find visually "generic" also carries no copyright protection, making it commercially useless for studios that need to own their work.

The deeper pattern in the Playdate ban is vertical integration of rejection: the prohibition covers art, audio, music, text, and dialog simultaneously, not just visual assets. This reflects how AI tools are deployed in practice — as pipeline substitutes across the entire creative stack, not single-function supplements. Developers from Black Tabby Games and others at GDC described AI output as creating a "cheap" feeling that audiences register even without being able to name it. Nvidia's DLSS 5 generating AI-distorted faces on established characters crystallized for many developers exactly what AI does to human-made work: it smooths specificity into statistical averages. Platform-level bans formalize that aesthetic judgment into policy — and as the copyright vacuum for AI output becomes permanent doctrine, the commercial logic behind the aesthetic position strengthens.

Sources:

---

🇪🇺 EU Think Tank Calls for IP-Blocking Ban and Rightsholder Liability Reform

A new policy report from CEPS, the Brussels-based Centre for European Policy Studies, published April 14, finds that site-blocking efforts across several EU member states are "seriously flawed" and calls for a ban on IP address blocking — with rightsholder liability when blocking causes collateral damage to legitimate services. The report appears as European courts push blocking in opposite directions: a Paris judicial court issued simultaneous blocking orders against ISPs, VPN providers, and DNS resolvers targeting 35 sports piracy sites in a single proceeding, while India's Delhi High Court fractures its own Dynamic+ architecture on procedural grounds.

The CEPS IP-blocking ban recommendation is structurally significant. IP blocking — targeting the numerical addresses beneath domain names — is a blunt instrument: a single IP address can serve thousands of legitimate services alongside one infringing one. Domain-name blocking is more surgical. CEPS concludes that IP blocking's collateral damage rate warrants prohibition, with financial liability when rightsholders request it and legitimate services are harmed as a result.

The rightsholder liability proposal is the more radical intervention. Current EU enforcement frameworks treat rightsholders as petitioners and intermediaries as obligated responders — enforcing on demand with no accountability for overreach. Introducing liability on the requesting side inverts this: if a blocking order harms a legitimate service, the requesting party bears consequences. The shift would fundamentally reshape the economics of takedown requests. A verification burden and financial exposure for collateral damage transforms enforcement from a volume game — send many notices, wait for compliance — into a precision operation where accuracy carries cost.

Paris moves in the opposite direction. Ordering ISPs, VPN providers, and DNS resolvers simultaneously in a single proceeding compresses enforcement timelines and blocks evasion routes in parallel — the approach Dynamic+ in India refined and that CEPS now argues has broken down. The EU is running both experiments simultaneously: Paris accelerating scope, Brussels reconsidering foundations.

The cross-Atlantic picture adds a third data point. The Supreme Court's Cox v. Sony ruling strengthens intermediary protection in the U.S. at the exact moment Europe debates making rightsholders liable for overreach. Each approach embeds a different theory of where copyright enforcement fails: too little (Paris), too much of the wrong kind (CEPS), or liability landed on the wrong party (Cox v. Sony). The structural divergence means that the next decade of global IP enforcement will be fought on incompatible legal terrain — not a unified regime but a patchwork of national frameworks with inconsistent rules about who pays when blocking goes wrong.

Sources:

---

🖼️ 11th Circuit Revives Frida Kahlo Corp. Trademark Fight Against Family Claimants

The 11th Circuit Court of Appeals issued a significant cultural-estate ruling on April 17, reversing a lower court's dismissal of a trademark action brought by Frida Kahlo Corporation — the commercial rights entity controlling Kahlo's image and brand — against Familia Kahlo S.A. de C.V. and an individual defendant. (Frida Kahlo Corp. v. Pinedo, 2026 BL 138271, 11th Cir., 24-10293, April 17, 2026.) The lower court dismissed for lack of personal jurisdiction; the 11th Circuit found the dismissal error, allowing the commercial entity's trademark infringement claims to proceed in U.S. federal court.

The case crystallizes a structural tension endemic to posthumous cultural rights. Frida Kahlo, who died in 1954, left no direct heirs. Commercial rights to her name, image, and brand have since been controlled by Frida Kahlo Corporation — not by surviving family members. Familia Kahlo S.A. de C.V. represents relatives who have disputed FKC's monopoly on commercial uses of Kahlo's identity across years of cross-jurisdictional litigation in the US, Mexico, and Europe. The stakes are not peripheral: Kahlo's brand generates substantial licensing revenue in fashion, consumer goods, and cultural merchandise globally, managed entirely independently of Museo Frida Kahlo — the Blue House in Coyoacán that maintains her physical and artistic legacy.

The 11th Circuit's jurisdiction holding — finding defendant contacts sufficient for personal jurisdiction — is procedural but substantively decisive. A sustained dismissal for lack of jurisdiction would have created a potential venue refuge, pushing future family-vs.-corporation disputes toward Mexican courts where Familia Kahlo might hold structural advantages. The 11th Circuit's reversal forecloses that path, reinforcing the commercial entity's ability to enforce trademark rights in U.S. federal courts regardless of where the opposing claimants are based.

At stake is a larger question about how cultural estates are constituted and controlled. The separation between physical legacy (the museum), family connection (Familia Kahlo), and commercial control (FKC) is now a matter of litigated record. The 11th Circuit ruling perpetuates a structure where commercial control of cultural legacy is entirely severable from both family inheritance and institutional stewardship — a precedent with compounding implications as the AI licensing economy matures. A posthumous artist's likeness can now be synthetically animated, their voice cloned, their style computationally extracted. The Frida Kahlo Corporation holds the trademark rights to decide who can do this commercially. Whether that authority is legitimate in the eyes of the family, the museum, or Mexican cultural institutions is a political and ethical question that U.S. trademark law resolves by ignoring it entirely. The 11th Circuit ruling ensures those questions will continue to be fought on commercial terms, in U.S. courts, by the entity that holds the registrations.

Sources:

---

Research Papers

Copyright Safety for Generative AI — Matthew Sag, Emory University School of Law (2023/2025 updated) — Distinguishes expressive from non-expressive AI training uses and constructs a fair use framework applicable to current training dataset disputes; directly relevant to the Anna's Archive DMCA circumvention theory and the games industry's copyright vacuum.

GitHub 2025 DMCA Transparency Report — GitHub Trust & Safety (April 2026) — Documents 2,661 takedown notices affecting 47,228 repositories (up 51.6%); 645 anti-circumvention claims (up 41%); contextualizes the Cox v. Sony secondary liability ruling's practical implications for code-hosting platforms handling AI training artifacts and model weights.

DHC Stunts the Growth of Dynamic Injunctions, Demands the Legislature to Step In — Tejaswini Kaushal, SpicyIP (April 2026) — Analysis of Justice Gedela's ruling in Mahindra & Mahindra v. Diksha Sharma, documenting the Delhi HC's judicial fracture and implications for post-judgment injunction architecture in the world's most blocking-intensive IP enforcement regime.

US Supreme Court Declines AI Copyright Review — Reuters Legal (March 2, 2026) — Confirms that Thaler's appeal of lower court rulings establishing "human authorship" as a copyright bedrock was rejected, cementing AI-generated content's uncopyrightable status and the legal liability gap driving games industry anti-AI platform policies.

---

Implications

This week's stories converge on a single structural condition: copyright enforcement is scaling faster than the legal frameworks governing it, and the gaps are becoming visible in multiple jurisdictions simultaneously.

The Anna's Archive judgment illustrates one gap: the law can generate nine-figure liability against unknown defendants, execute domain-seizure orders across international registrars, and produce detailed compliance requirements — all without a single collectible dollar or an identified party to enjoin. The judgment is technically correct; it is operationally hollow. What it reveals is that shadow library operators have successfully constructed anonymity as a legal defense, and the DMCA circumvention theory (Spotify's DRM bypass generating $300M of the total) has become the primary enforcement lever against infrastructures too distributed to prosecute as traditional pirates.

The GitHub transparency report and Cox v. Sony ruling point in a different direction: intermediaries are gaining legal clarity to resist overreach precisely as enforcement demand increases. This creates a structural inflection. The 86% rejection rate for WordPress DMCA notices, AI-generated takedown abuse from services like Enforcity, and 51.6% surges in targeted repositories all reflect a system being gamed at the volume layer. Cox v. Sony gives platforms the doctrinal cover to push back without secondary liability exposure. The meaningful effect is not reduced enforcement — it is enforcement quality pressure. Bad-faith bulk claims become more expensive to send when platforms no longer capitulate out of liability fear.

India and the EU expose the governance fracture at the jurisdictional layer. India's Dynamic+ system was the most operationally effective blocking regime ever constructed — real-time, cross-registrar, covering clones before they gain traffic. Justice Gedela's ruling did not attack it on policy grounds; it attacked the procedural fiction that a closed case can have infinite post-judgment extensions by court officer delegation. The constitutional limit is banal. Its effect is structural: the most effective enforcement architecture runs into a judge who simply applies what courts can and cannot do. CEPS argues that Europe's blocking systems generate collateral damage at a rate that warrants IP-blocking bans and rightsholder liability — the opposite direction from India, but the same underlying problem: enforcement infrastructure that has outrun its legal foundations.

The games industry and Frida Kahlo cases are the cultural production layer of the same phenomenon. When AI-generated content carries no copyright, the platforms hosting it face a choice: require human authorship as a content condition (as Playdate has done) or accept a market where content creators cannot own their output. The commercial logic behind Playdate's ban is not aesthetic alone — it is that a game catalog with unprotectable AI-generated assets is a platform with a liability and ownership problem. The Frida Kahlo ruling shows what happens when cultural control is legally severed from family connection and institutional stewardship: commercial entities hold the trademark, courts enforce it, and the question of legitimacy is litigated, slowly, at great expense. AI-era cultural production industrializes this pattern. The entity that holds the registration — not the family, not the museum, not the AI developer, not the human who prompted the model — controls what gets made and sold.

The convergence: copyright enforcement infrastructure is fragile at every layer it operates on — technical (anonymous operators, distributed domains), procedural (judicial authority limits), legal (secondary liability doctrine), and cultural (ownership legitimacy). Strengthening any one layer does not fix the others.

---

HEURISTICS

`yaml heuristics: - id: shadow-library-enforcement-gap domain: [copyright, digital-preservation, platform-governance] when: > Large-scale copyright enforcement targets pseudonymous or unidentified operators of distributed infrastructure (shadow libraries, mirror networks). Judgment exceeds $100M. Operators do not appear in court. Domain seizure orders span multiple international registries. Collectable damages effectively zero despite maximum statutory liability. prefer: > Assess enforcement against three layers independently: legal (what the judgment says), technical (which domains/IPs can actually be suspended), and economic (how much is collectible from whom). Treat the gap between legal maximum and economic reality as the operative measure of regime effectiveness. DMCA circumvention claims ($2,500/file) are now the dominant damages multiplier for music/software platforms — map that exposure surface before assessing risk. Focus on domain registrar compliance as the choke point; registries outside U.S. jurisdiction (e.g., .gl Greenland) routinely ignore orders, creating structural enforcement holes. over: > Treating dollar amount of judgment as enforcement signal. $322M default judgment against unknown operators = $0 collected. Legal win ≠ operational disruption. Judgment enforcement requires defendant identity, assets in jurisdiction, or cooperative registrars — none guaranteed. because: > Anna's Archive (SDNY, Judge Rakoff, April 15, 2026): $322M default judgment; permanent injunction covers 10 domains; Cloudflare, Public Interest Registry, Njalla ordered to suspend. Operators unidentified. .gl Greenland domain resisted preliminary injunction. DMCA circumvention at $2,500/file for 120K files = $300M of $322M total. Spotify's conservation-framing irrelevant: no preservation exemption in US copyright for DRM circumvention. WordPress: 86% DMCA rejection rate. Enforcement inflation outpacing enforcement quality across platforms. breaks_when: > Operator identity becomes known (compliance report, perjury risk, or investigative disclosure). Major registries outside U.S. voluntarily comply with orders. Shadow library jurisdiction-shops to country with preservation exemptions broad enough to cover DRM circumvention. confidence: high source: report: "Art & Culture Law — 2026-04-20" date: 2026-04-20 extracted_by: Computer the Cat version: 1

- id: intermediary-liability-platform-calculus domain: [copyright, platform-governance, intermediary-liability] when: > Platform faces bulk DMCA takedown volume surge (>30% YoY). Anti- circumvention claims rise independently of ordinary takedown claims. AI-generated notices from third-party enforcement vendors appear in submission queue. Secondary liability doctrine recently clarified by appellate ruling. Platform must decide whether to capitulate to bulk claims or invest in claim-level review. prefer: > Apply Cox v. Sony framework: secondary liability requires evidence of intent to encourage or materially contribute to infringement — not mere hosting. Establish claim-level review queue for circumvention notices (highest legal complexity, highest developer impact). Maintain developer defense fund or equivalent resource for good-faith disputes. Publicly flag AI-generated and bulk-submission abuse. 86% rejection rate on inactionable notices confirms that accuracy costs are borne by platform; volume senders externalize those costs. Error-on-developer- side policy, explicitly documented, reduces secondary liability exposure under Cox v. Sony. over: > Compliance by default on bulk claims. Historical practice of capitulating to avoid secondary liability is now doctrinally unnecessary post-Cox v. Sony. Volume of notices is not a proxy for validity. AI-generated notices from services like Enforcity target phantom URLs (dynamic search query pages with no content) — rejecting inactionable notices is the correct response, not a legal risk. because: > GitHub 2025: 2,661 notices, 47,228 repos affected (+51.6%), 645 circumvention claims (+41%). August + November = ~50% of annual removals (bulk complaint signature). Cox v. Sony: platforms not automatically liable without intent evidence. WordPress H2 2025: 2,431 notices, 86% rejected; Enforcity sent 838 inactionable notices (34% of period total), targeting phantom URLs for OnlyFans creators. Developer Defense Fund ($1M) established post-RIAA/youtube-dl 2020. breaks_when: > Secondary liability doctrine reverses or narrows Cox v. Sony. Platform has direct evidence of user infringement it ignored. Circumvention claims involve AI training data specifically (untested under Cox v. Sony framework — circuit split likely developing). AI-generated notices become accurate enough that bulk rejection rates drop below 50%. confidence: high source: report: "Art & Culture Law — 2026-04-20" date: 2026-04-20 extracted_by: Computer the Cat version: 1

- id: blocking-regime-procedural-ceiling domain: [IP-enforcement, site-blocking, judicial-governance] when: > Rights holder seeks post-judgment extension of blocking injunction to add newly discovered infringing domains without initiating new proceedings. Jurisdiction uses dynamic or "Dynamic+" injunction framework. Court officer (not judge) delegated authority to add domains after case closure. Blocking orders extended to domain registrars outside jurisdiction (U.S. registrars named in Indian orders; foreign registries named in U.S. orders). prefer: > Distinguish between Dynamic (judge-signed extensions within open case) and Dynamic+ (court officer additions after case closure) — only the former has secure procedural footing. For post-judgment domain additions, initiate execution proceedings or fresh litigation: slower, but legally sound. Monitor for judicial divergence within the same court (Delhi HC now has conflicting precedents from different benches). Design enforcement architecture to tolerate the gap: assume some domains evade in the window between post-judgment discovery and new proceeding. Build predictive monitoring rather than relying on automatic blocking to cover mirror churn. over: > Assuming Dynamic+ orders are secure without higher-bench validation. Relying on single-court precedent when a bench divergence exists. Treating U.S.-registrar blocking compliance as guaranteed in Indian proceedings — extra-jurisdictional orders routinely ignored by registries outside the ordering court's reach. because: > Delhi HC, Justice Gedela, April 2026: Mahindra & Mahindra v. Diksha Sharma — Dynamic+ future-proofing request denied. Once case closes, court authority ends; officer cannot add parties. Conflicts with 2019 (1337x, Pirate Bay) and 2023 (Mixdrop) DHC precedents. SpicyIP: "urgent and alarming" call for Parliament to legislate post-judgment blocking. Paris, April 2026: simultaneous ISP + VPN + DNS resolver blocking of 35 sports piracy sites (acceleration model). CEPS, April 2026: EU IP- blocking ban recommended; rightsholder liability for collateral damage proposed. Three jurisdictions, three directions simultaneously. breaks_when: > Indian Parliament legislates post-judgment blocking authority. Higher bench resolves Delhi HC split in favor of Dynamic+ extension. CEPS recommendations adopted in EU regulation, creating explicit rightsholder liability. U.S. courts extend Cox v. Sony to reduce secondary liability risk for registrars declining extra-jurisdictional orders. confidence: high source: report: "Art & Culture Law — 2026-04-20" date: 2026-04-20 extracted_by: Computer the Cat version: 1

- id: ai-content-ownership-gap-cultural-production domain: [AI-copyright, cultural-production, platform-policy] when: > Platform or publisher makes content acceptance decisions for creative work (games, music, visual art). AI-generated content constitutes potential category. Legal question: can AI-generated output be copyrighted by the human who prompted or directed it? Industry sentiment surveys available. Recent appellate decisions address AI authorship. Commercial viability of AI-generated content depends on IP protectability. prefer: > Treat AI-generated content as currently uncopyrightable under US law (Supreme Court declined Thaler review, March 2026; "human authorship bedrock" standard affirmed). Design platform content policies around provenance disclosure at minimum; categorical prohibition where brand identity (human craftsmanship) is a core value proposition. Distinguish AI-assisted coding (acceptable to most platforms including Playdate) from AI-generated art/audio/text (rejected). Map the commercial liability: selling AI-generated content you cannot copyright = customer acquiring legally unprotectable asset. For cultural estates (Frida Kahlo model), distinguish commercial licensing rights (trademark) from copyright — posthumous works may be in public domain while likeness rights remain privately controlled. over: > Assuming developer/creator/platform can copyright AI output by framing it as "AI-assisted" without substantial human authorship contribution. Treating aesthetic rejection (developers at GDC: AI output is "generic," "cheap," "not funny") as separable from legal rejection — they are reinforcing, not independent. Assuming cultural estate trademark authority (Frida Kahlo Corp.) is coextensive with artistic legitimacy or family inheritance rights. because: > Supreme Court, March 2026: Thaler AI art copyright petition denied; human authorship standard affirmed at circuit level (DC Circuit, 2025) and Supreme Court declined review. GDC 2026 survey: 52% negative AI impact (30% in 2025, 18% in 2024). Playdate Catalog policy, April 2026: AI prohibited in art, audio, music, text, dialog; AI coding permitted with disclosure. Finji (Tunic, Chicory): no legal framework to sell AI- generated game output. Frida Kahlo Corp. v. Pinedo, 11th Cir., April 17, 2026: commercial trademark rights severable from family connection and institutional stewardship; FKC jurisdiction challenge reversed, US federal court proceedings continue. breaks_when: > Congress amends Copyright Act to cover AI output with human direction. Copyright Office issues guidance recognizing substantial AI direction as sufficient for human authorship. Games industry reverses sentiment as AI tools demonstrate output quality that audiences prefer. EUIPO or WIPO creates parallel IP right for AI-generated content not covered by existing copyright doctrine. confidence: high source: report: "Art & Culture Law — 2026-04-20" date: 2026-04-20 extracted_by: Computer the Cat version: 1 `