A free-expression group's AI paper is not a copyright brief

The writer's problem

A writer trying to place generative AI's risks in context, beyond the copyright infringement suits filed by individual authors, needs a source addressing free expression and creative labor as such. On 31 July 2023, PEN America published Speech in the Machine: Generative AI's Implications for Free Expression, a 36-page white paper distinct from any single lawsuit's claims.

What the documents show

The accompanying press release states the paper warns that generative AI could narrow the range of creative works available and undercut authenticity in public discourse, while also noting opportunities for new forms of expression. PEN America's chief program officer for free expression is quoted saying the organization must safeguard the uniquely human spark of inspiration that drives the creative process. The paper's recommendations to policymakers, per the release, urge combining rights-based and risk-based regulatory approaches rather than treating them as a forced choice, and call for transparency, researcher access to AI systems, and effective remedies where AI drives automated decisions. These are PEN America's own stated positions and principles, offered as guidance to policymakers and industry, not findings from a controlled study or a court's ruling.

The editorial choice

PEN America explicitly connects this paper to its earlier disinformation work, citing its 2017 Faking News report and 2019 Losing the News report as context for treating generative AI as an extension of, not a break from, threats it had already been tracking. A writer citing the white paper should present its principles as one advocacy organization's framework for policymakers, not as a consensus position across the free-expression field.

What stays with the author

The paper is a policy document aimed at regulators and platforms; it does not resolve how any individual writer should respond to generative AI in their own work, and PEN America's press materials do not claim it does. Judging whether a specific AI use threatens a specific piece of creative work remains a case-by-case editorial judgment the white paper's general principles do not make for a reader.

• Is a cited AI risk here a PEN America principle for policymakers, or a claim about a specific product?
• Does the source distinguish free-expression risk from the separate question of copyright infringement?
• What would PEN America's own recommended transparency and researcher-access standards require of the tool in question?

Speech in the Machine reads as a mission statement for regulation debates rather than a manual for any one writer's manuscript, and it is most useful cited as exactly that.