THE FEAR IS RATIONAL.
In 2025–26, every major holding company shed thousands of jobs while executives told investors AI would let them do more with fewer people. These aren't projections. They already happened.
None of those decisions involved the people doing the work. That's the actual problem: not the technology, the total absence of your voice when it's deployed.
THREE GROUPS ALREADY WON THIS FIGHT.
Screenwriters, performers, and game workers all faced the same question you do, and all three walked away with enforceable AI language in a contract. Here's exactly what they got.
WHAT THE WGA WON
2023 strike → protections preserved & extended in the 2026 MBA (the Minimum Basic Agreement, the WGA's master contract), ratified April 2026 with 90.38% member approval
AI cannot write or rewrite literary material, and AI-generated text is not "source material" that undercuts a writer's credit or pay.
No writer can be required to use AI tools as a condition of employment.
If a company hands a writer AI-generated material, it must disclose that up front.
New in the 2026 MBA: if a company licenses writers' scripts to train a commercial generative-AI system, it must give the Guild written notice, and the Guild can demand discussion of the license, including remuneration for writers.
WHAT SAG-AFTRA WON
2025 Interactive Media Agreement, ratified 95.04% after an 11-month video-game strike
Written consent required before a company can create or use a digital replica of a performer's voice, face, or movement.
Separate, negotiated compensation for digital replica use. Replication is paid work, not a freebie.
Disclosure requirements: performers must be told when and how replicas will be used.
Performers can suspend consent for the generation of new AI material during a strike. Consent is leverage, not a one-time signature.
WHAT ZENIMAX WORKERS WON
ZeniMax Workers United-CWA first contract, June 2025: Microsoft's first US union contract
AI-use protections: the employer must give prior notice of AI deployments that affect union members, and negotiate those impacts at the union's request.
A crediting policy so QA workers are named in the games they help ship.
13.5% across-the-board raises plus new minimum salary floors.
A grievance process and protection against arbitrary dismissal: the enforcement machinery that makes the AI language real.
FIVE DEMANDS, TRANSLATED FOR AD CREATIVE.
Every one of these maps directly to language already won at a bargaining table somewhere. This isn't a utopian wishlist. It's precedented contract text, translated into what it would mean at an agency.
You're a bargaining unit of one, and the same five clauses translate directly into terms for your next SOW: no training on your work without written consent, disclosure when you're handed AI-generated inputs, no rate cuts for tool use. Put them in your own contracts. Here's the freelancer playbook.
NONE OF THIS IS AVAILABLE TO YOU ALONE.
As an individual at-will employee (they can let you go almost anytime, for almost any reason), you cannot make your employer:
- disclose AI-generated material
- pay you when the model trains on your work
- promise that adopting a tool won't cut your rate
You can ask. They can say no. And tomorrow they can hand you a severance packet generated by the thing that replaced you.
Every protection on this page exists for exactly one reason: workers bargained for it, and the result got written into an enforceable contract. Writers struck for it. Performers struck for it. Game workers spent two years at the table for it.
Staff need a unit to win it; freelancers can write the same five clauses into their own contracts, per the callout above. The playbook exists. The precedents exist. What's missing is ad creatives running it.