Studio Ghibli, Bandai Namco, Square Enix Demand OpenAI to Stop Using Their IP

Anti-piracy analogy & data harvesting

• Many compare AI training on copyrighted works to classic piracy: “downloading content for AI training is stealing.”
• Others argue the “you wouldn’t steal a DVD/car” analogy is weak because digital copies have zero marginal cost and harm is indirect or market-dependent.
• Some highlight the irony that past anti-piracy campaigns themselves used infringing material, underscoring the complexity and hypocrisy around IP.

Ads, attention, and what counts as “payment”

• One side claims pervasive advertising “steals” time, attention, mental health, and device resources.
• Counterargument: viewing ads is a voluntary payment for a service; you can refuse by not using the service or by paying directly.
• Tension appears when companies call ad-blocking “theft” while asserting ads are a fair exchange.

Transformative use, scale, and AI vs humans

• Broad agreement that AI pushes the limits of “transformative use” doctrines: the law never anticipated systems that ingest everything and output in any style at scale.
• Some insist embedding works in vector spaces is not meaningfully transformative; others say we don’t fully understand human creativity either, so process-based distinctions may be shaky.
• A recurring theme: scale and automation change the ethical and legal calculus even if AI “learned” similarly to humans.

Style, copyright, and legality

• Several comments stress that styles are generally not copyrightable; specific characters, plots, and compositions are.
• Disagreement over whether painting in “Ghibli style” is infringement or simply fair use / non-actionable inspiration, especially for non-commercial personal work.
• Others argue that when a commercial product (e.g., OpenAI) systematically enables Ghibli-like output and sells access, it crosses into direct competition and likely infringement.

Artist livelihoods and cultural impact

• Strong concern that AI undermines artists’ ability to earn a living by cheaply cloning styles built over lifetimes.
• Some say this is akin to “corporate piracy” or exploitation; others counter that art has always been copied and that business models—not art itself—must adapt.
• A few take a hard line: many artists may have to “get a stronger business” or leave the profession; others warn that losing working artists degrades culture, critical thinking, and “human” entertainment.

Enforcement, jurisdictions, and future models

• Debate over whether model training is currently illegal; some say it’s clearly willful commercial infringement, others assert training is lawful but outputs may infringe.
• Non-US perspectives note that many countries lack broad fair-use concepts; examples from Japan suggest that even using Ghibli-like AI images commercially could trigger counterfeiting laws.
• Some expect outcomes analogous to Napster (banned) vs YouTube (licensed); others predict large payouts, “firewalls” around national IP, or robots.txt-style opt-outs becoming mandatory.

Fairness, double standards, and what the law ought to be

• Several point out a double standard: individuals happily pirate but condemn OpenAI; big tech invokes “fair use” aggressively while defending its own IP.
• Others emphasize that beyond what’s currently legal, society must decide whether it’s fair for a few companies to appropriate “the treasure of humanity” without consent or attribution.
• There’s no consensus: views range from “end fair-use harvesting” to “I hate copyright more than I hate AI companies,” with many admitting they are genuinely torn.