Becerra Campaign Policy Flagged as AI-Generated
A 1,600-word film policy statement from the gubernatorial candidate tested at 100% machine-authored.
Updated on Oct. 14, 2026 in Artificial Intelligence
Live Poll
Does the use of artificial intelligence to draft political policy statements undermine your trust in candidates?
The California gubernatorial campaign of Xavier Becerra released a 1,600-word film policy statement that was identified as 100% AI-generated by the Pangram detection tool. State disclosure laws currently mandate labeling for audio, video, and imagery, but text is excluded from these requirements.
Why it matters
The incident highlights a growing gap between regulatory disclosure requirements and the widespread integration of generative models into political communications. As candidates utilize these tools, voters face increasing difficulty in verifying the human authorship of detailed policy proposals.
The Pangram detection tool evaluated the 1,600-word document and assigned it a 100% probability of being AI-authored. The tool is calibrated with a false positive rate of 1 in 10,000 for standard text and 0.0% for political speeches.
The players
Xavier Becerra
Gubernatorial candidate currently holding a 20-point lead in polls and a 96% win probability in prediction markets.
Rob Bonta
Attorney General who intends to sponsor a bill to eliminate the $750 million cap on state film and TV production incentives.
The details
The Pangram tool functions by analyzing linguistic patterns and statistical probabilities in prose to identify sequences typical of large language models—algorithms trained on massive datasets to predict the next word in a sequence. While the tool identified the document as synthetic, California law currently requires disclosure only for manipulated audio, video, or images. This means that while a deepfake video or audio clip would require a mandated disclosure label, textual policy documents remain outside the scope of current state transparency statutes.
Timeline
Monday: Becerra discussed film policy at an event in Glendale.
Last month: The Economist analyzed parliamentary speeches for AI use.
Early next year: State lawmakers plan to strengthen the film incentive program.
The Tech Race
This development follows a documented trend identified by The Economist, which found that 10% of words in British parliamentary speeches are now drafted using AI tools. It marks a transition toward automated policy generation within California's high-stakes gubernatorial environment.
Voters and policy observers should note that state-mandated AI disclosure rules currently exclude text, meaning AI-written policy documents will not carry mandatory labels. This leaves the burden of verification on researchers using third-party tools like Pangram until potential legislative updates occur.
The takeaway
The event highlights that as LLM-assisted drafting becomes standard in state politics, existing disclosure laws remain limited to visual and auditory media. Observers should track the upcoming legislative efforts in Sacramento early next year to see if transparency requirements expand to include generative text.
Further reading
For more on how language models are evolving, see Artificial Intelligence.
Live Poll
Does the use of artificial intelligence to draft political policy statements undermine your trust in candidates?
