SAG-AFTRA's AI guardrails attach to the employer that deploys the tool

The scoping line is the part a product team can use. Duncan Crabtree-Ireland, SAG-AFTRA's chief negotiator, told The Verge the union "bargained for and achieved enforceable, meaningful guardrails around how this technology can be implemented by the major employers in our industry" ^[4]. He said those terms do not include companies like OpenAI, Anthropic or Google directly, but do encompass "all of the companies who have built up businesses around the creation of professional level entertainment content in the United States" ^[5]. The obligation follows the employer who deploys a tool.

Now put that beside the list of firms asking for a pause. Executives at Anthropic, OpenAI, Google DeepMind and X have called for an industry-wide slowdown in development ^[10], and Anthropic alignment lead Evan Hubinger said the technology could "kill all humans" at some point "within the next decade" ^[9]. Those are the same companies Crabtree-Ireland placed outside the union's bargained perimeter ^[16]. President Donald Trump has dismissed the concerns as overreactions, while some other US politicians have felt pressure to legislate ^[11].

Crabtree-Ireland's own position runs both ways. He said people "should be concerned about the broader societal and social implications of AI," and also thinks some of the extinction-level warnings could be hyperbole ^[6]. Hearing it from "the CEOs of these companies or the billionaires who've made their billions off of doing the exact thing that they're now warning us might not be good" has left him unsure "what to trust or what to believe," he said ^[7]. He credited Jacob Coxon, the Anthropic researcher who quit to sound the alarm, with getting "traction with the public" ^[8].

A media team's AI review covers the provider's safety posture, the model card, the eval numbers in the vendor deck. The person who rolls the tool out answers for a crew member who believes the tool was used against them. The Verge reports that the unions' fight for strict rules did not stop the entertainment industry from embracing AI, and that it gave Hollywood's laborers a way to fight back when they feel the tools are being used to harm them ^[13]. The Verge also reports that production AI does not appear as existentially dangerous as the systems behind the recent whistleblowing, while models that generate digital facsimiles of actors and replicate traditional filmmaking tools threaten some human jobs ^[12].

So two conditions decide where a rollout sits. First, the deploying entity is one of the US professional content businesses Crabtree-Ireland described as covered ^[5]. Second, the tool's output touches a performer's likeness or takes over work a covered crew member does ^[12]. Two yeses put the rollout inside terms the unions bargained during the 2023 strikes ^[14], and the safety literature coming out of the labs will not tell you what those terms require. One yes, and the argument is about procurement. The studios named in The Verge's outreach have stayed quiet ^[1].