The latest AI announcements from Blizzard, ElevenLabs, the Australian government, Amazon Prime Video, and the US Department of Justice.
VOL. 2 SEPTEMBER 2026
INFO-TECH AI INSIGHTS
Creative Intelligence
Featuring AI news, best practices, tools, and insights to help creative professionals strategize, experiment, create, produce, and responsibly integrate AI while protecting their creative rights, careers, and intellectual property.
AI IN THE ARTS
Blizzard's contract makes AI a bargaining matter
On September 9, 2026, nearly 1,900 Blizzard Entertainment employees ratified their first union contract with Microsoft after two years of negotiation, covering the World of Warcraft, Overwatch, Diablo, and Hearthstone teams. It provides wage increases, a three-day hybrid workweek, four weeks of supplementary severance, and an industry-first entitlement to recall into open positions within 14 months of a layoff. It also obliges Microsoft to bargain with the union before deploying generative AI tools that materially affect represented roles.
ANALYST ANALYSIS
This is the video game industry's first major agreement requiring an employer to bargain over generative AI before deployment rather than after the fact. It does not prohibit AI, but instead designates any workflow change that materially affects represented positions as a negotiable event, which gives nearly 1,900 employees influence before new tools arrive. That leverage matters because the contract follows Xbox's elimination of 1,600 roles with a further 1,600 anticipated, a sequence that makes AI adoption a labor-relations question as much as a technical one. Comparable consultation obligations should be expected across unionized creative sectors.
ElevenLabs bets on licensed catalog over scraped audio
Read the ElevenLabs announcement
On September 10, 2026, Universal Music Group and the AI audio developer ElevenLabs announced a multi-year licensing agreement, representing ElevenLabs' first partnership with a major record label. The first product under the agreement is an AI music creation platform built on licensed catalog and artist consent, enabling audiences to generate remixes, mashups, and personalized vocal renditions of those artists' recordings. The parties intend to develop further AI audio products for artists and songwriters.
ANALYST ANALYSIS
The agreement couples licensing with joint product development, so the consumer platform draws exclusively from licensed catalog and from artists who have elected to participate. That structure reflects a deliberate wager, as ElevenLabs is betting that rights-cleared repertoire rather than scraped material will prove the more durable foundation for an AI audio enterprise. Notably, the licensed platform is to be kept distinct from ElevenLabs' existing music products, a two-track approach in which rights-cleared material underpins the label partnership while the established models operate separately. By compensating artists and securing rights before building product, the partnership offers rights holders an early framework for monetizing AI capability rather than pursuing litigation alone.
Leaked Australian government proposal shifts the burden from AI companies to creators
On September 15, 2026, Australian government documents seen by the Australian Broadcasting Corporation set out a proposal that would permit AI companies such as OpenAI and Anthropic to train on online content without first obtaining creators' consent. Two approaches were presented. Under the first, companies pay a central body that distributes funds to registered copyright owners. Under the second, they need only conclude agreements with a prescribed number of rights holders. Both would render training opt-out by default.
ANALYST ANALYSIS
Two proposals are under consideration, and the distinction matters for creators. The first establishes a compensation pool, with AI companies paying a central body that distributes funds to registered copyright owners. The second requires only that a company conclude agreements with a prescribed number of rights holders, so no compensation would reach creators beyond them. Under either, training is opt-out by default, so silence constitutes consent and the burden of protection falls to creators, who must lodge and relodge opt-outs indefinitely and without compensation across every site displaying their work, including instances where their work was uploaded by third parties. Creators should check whether their guild or collecting society is party to these consultations and seek opt-out and AI-training provisions in new contracts.
AI-assisted dubbing on Prime Video changes what audiences see
On September 9, 2026, Amazon's Prime Video introduced AI-assisted lip-sync dubbing on Maxton Hall, its most-viewed international original series. Professional voice actors continue to record the dubbed dialogue, while AI and visual effects adjust the performers' on-screen mouth movements to correspond with that audio, eliminating the mismatch conventional dubbing produces. It is available on seasons 1 and 2 in English, will extend to Season 3 in December, and will reach additional titles.
ANALYST ANALYSIS
Amazon is not displacing voice actors, as they continue to record the dubbed dialogue while AI modifies only the performers' on-screen mouth movements so that the two align. Deploying the capability first on Prime's most viewed international original suggests a controlled trial ahead of broader rollout. Even so, it raises a question studios must eventually address contractually, namely the extent to which a performance may be altered by AI. Retaining the human performance in the audio while modifying it visually draws a distinction that warrants explicit definition in future agreements.
Gadot's AI contract is becoming a reference point for SAG-AFTRA
On September 1, 2026, Gal Gadot disclosed that she spent six months negotiating her contract for Doug Liman's AI-driven thriller Bitcoin, securing terms permitting AI to be applied to sets and lighting but prohibiting any alteration of her performance. She called it the most protracted negotiation of her career and noted that counsel for SAG-AFTRA approached her legal team for guidance on revising the guild's own AI provisions.
ANALYST ANALYSIS
The significance of Gadot's six-month negotiation went well beyond her own engagement, since SAG-AFTRA subsequently sought guidance from her counsel in revising its AI safeguards. Her stipulation was narrow but consequential, permitting AI on sets and lighting while requiring that her performance remain unaltered, and that distinction is emerging as the operative line performers and unions assert. Her own posture is equally instructive, as she concedes she is apprehensive about the technology yet has chosen to learn it, observing that the train has left the station. For creative professionals, the implication is that AI is unlikely to recede, so the more proactive approach is to negotiate firm limits on its use while acquiring enough fluency to remain competitive.
Washington signals support for AI developers in court
Read the Hollywood Reporter article
On September 1, 2026, the US Department of Justice filed a statement of interest supporting OpenAI in its copyright litigation with the New York Times, arguing that training large language models on copyrighted text is "extraordinarily transformative" fair use. It is the federal government's first formal intervention in this wave of AI copyright litigation. The Times characterized the filing as Washington "siding with a handful of trillion-dollar AI companies at the expense of the countless American creators whose work they stole."
ANALYST ANALYSIS
In its first intervention in this litigation, the Department of Justice advised a New York federal court that training large language models on copyrighted text is "extraordinarily transformative" and should therefore qualify as fair use, cautioning that restricting the practice would impair US competitiveness and national security. The filing carries no binding authority, as the court will rule independently, but it does indicate the direction of the US government's thinking on a question that continues to divide the courts. It should therefore be read as a meaningful, if unresolved, signal that US policy currently favors developers over rights holders.
Hollywood veterans sell craft expertise as AI training data
On August 22, 2026, The Guardian reported that established and in certain instances award-winning Hollywood writers, directors, and producers are being paid $12 to $200 per hour on gig platforms including Mercor, Micro1, and Handshake to instruct AI models in the granular judgment their occupations depend upon, from constructing a shoot schedule to labeling voices in raw audio. Motion picture and sound-recording employment has declined 28% from its 2022 peak, and many participants characterize the work with resignation rather than enthusiasm.
ANALYST ANALYSIS
This arrangement converts decades of accumulated craft expertise into reusable training data at rates far below what the same judgment commands in production work, and it does so permanently, since a model retains the instruction long after the contract concludes. For many it is a difficult decision, since the contraction in production work narrows the alternatives, and these occupations are structured around project-based engagements that afford little income continuity between them. Creators weighing these offers should establish what their contributions will be used to build and whether the engagement is disclosed, while organizations recruiting displaced specialists should anticipate lasting reputational and retention consequences.
Promise backs independent filmmaking over big-studio budgets
On August 16, 2026, The Guardian reported from Promise, an AI studio backed by Google and Disney adjacent to Sony's historic Culver City lot, during production of the AI-enabled horror feature "Touch Grass." Performers act on camera while AI generates the surrounding environments and effects, rendered live on the director's monitor. Visual effects veteran Joel Hynek likened the transition to the shift from optical to digital filmmaking, while others cautioned that positions will be displaced before the technology matures.
ANALYST ANALYSIS
At Promise, performers act with AI-generated environments and effects rendered live on the director's monitor. Opinion on what that represents is divided, as Academy Award-winning visual effects artist Joel Hynek considers the transition comparable to the move from optical to digital filmmaking, while directors, performers, and legislators warn that positions will be eliminated before the technology matures. Executives at these studios estimate hybrid production runs from 20% to 50% cheaper, which loosens the hold studio financing has over what gets made. The opportunity is genuine, since the technology empowers anyone with a talent and a discipline for storytelling, but the open question is whether that gain reaches creators or accrues to the few technology companies supplying the models.
CREATIVE AI RESOURCES
CREATIVE AI TOOLKIT:
Tools and platforms to create and protect your work with AI.
Fairly Trained – Certifies generative AI models trained exclusively on licensed or public-domain data.
Glaze & Nightshade (University of Chicago) – Free tools that shield artwork from unauthorized AI scraping and style mimicry.
LEGAL & IP WATCH:
Stay current on copyright law, litigation, and your rights.
EUIPO – European Union Intellectual Property Office – Official EU guidance on copyright, AI training, and preserving creators' rights.
US Copyright Office – AI & Copyright Portal – Official hub tracking copyright law, guidance, and rulings on AI-generated works.
SAG-AFTRA – AI Resources – Contract language and rights guidance for performers navigating AI use.
The Authors Guild – AI & Advocacy Center – Opt-out contract clauses and advocacy guidance for writers.
LEARNING CORNER:
Courses and guides to build your AI literacy.
The Ivors Academy – Knowledge Toolkit – Guides on AI, music rights, and royalty protection for songwriters.
Creative Commons – AI & the Commons – Explainers on open licensing, public domain, and generative AI for creators.
Partnership on AI – Responsible Practices for Synthetic Media – Best-practice guidance for creators experimenting with synthetic content.
INFO-TECH AI RESOURCES
Artificial Intelligence Research Center
AI Workforce Development Program
CREATIVE INTELLIGENCE EDITORIAL STAFF
Dana Cho – Info-Tech Research Specialist and Editor-in-Chief
Bill Wong – Info-Tech AI Research Fellow
