Move Faster or Go Unheard

Happy Monday,

There’s a three-week problem in Washington, and I’m building the solution.

Grassroots campaigns are effective but incredibly challenging to orchestrate. Mobilizing advocates on-message takes weeks. By the time your advocates are ready, the issue has already moved.

Social media, AI content, and the political incentives that follow are narrowing the window to influence any issue. Relevancy is the currency of social media. Once an issue goes cold, algorithms yield no share of voice. Traditional grassroots tactics leave clients at a disadvantage in fast moving issue spaces.

If you’ve been reading this newsletter, you know that new technologies are transforming opinion shaping in Washington. I’m excited to share that I am building a new firm that will deliver AI native public relations to shape public opinion at scale.

A series of happy accidents led me to this point. By chance, I got the ChatGPT beta just before the public launch which gave me—a journalism student—the skills to start an AI consultancy. Later on, circumstances led me to stand against injustice on campus. I spearheaded a grassroots campaign where we leveraged AI to achieve our goals with a small, inexperienced team working under extreme time constraints. I saw a glimpse of what AI could bring to opinion shaping and decided to go to Washington.

Working in PR taught me that the constraint is not the strategy or messaging, but the weeks it takes to deploy it.

If you or someone you know are looking for ways to move faster and shape conversations directly inside the feed, I’d welcome an introduction.

Now, onwards to our signals.

The Signals

Six signals of how AI is changing influence in Washington.

1. Anthropic Publishes Political Even-Handedness Audit for Claude

Anthropic released a detailed report on how Claude is trained and evaluated for political even-handedness. The company scored Claude Sonnet 4.5 at 95% even-handedness, matching Google’s Gemini 2.5 Pro (97%) and Elon Musk’s Grok 4 (96%), while outperforming GPT-5 (89%) and Meta’s Llama 4 (66%).

The move comes after intense political pressure. President Trump signed an executive order barring federal agencies from procuring AI systems that “sacrifice truthfulness to ideological agendas.” David Sacks, the White House’s AI czar, has publicly accused Anthropic of pushing liberal ideology. Anthropic responded by open-sourcing their evaluation methodology and publishing competitive benchmarks.

The report tests models using thousands of prompts covering hundreds of political stances, grading responses on even-handedness, inclusion of opposing perspectives, and refusal rates. Anthropic has rewritten Claude’s system prompt to avoid unsolicited political opinions, refrain from persuasive rhetoric, and use neutral terminology.

Something to think about: For Washington operators, AI as the new mediator of information means it is a new persuasion surface. If AI systems publish political-bias dashboards, influence work becomes partly about shaping the informational environment those models learn from. Issue content and high-authority explanatory material all become inputs that affect how Claude answers constituent questions. It’s a good time to begin systematic GEO testing of your core issues: How does the AI describe your client’s bill, industry, or controversy? What sources does it pull from? Yes, this sounds like a broken record, but you can’t influence humans at scale if you don’t influence their default AI answer engines.

Read more: Anthropic Political Even-handedness

2. Marble Introduces Spatial World Models—A New Narrative Surface

World Labs unveiled Marble, a spatial world model that generates explorable 3D environments from text prompts, treating physical space itself as a new programmable medium. Marble argues that spatial worlds will soon be as easy to generate as images or video. If you don’t understand what that means—and it’s a little out there—I recommend you try their generous free plan.

The platform offers multiple input types—text, images, videos, panoramas, or 3D layouts—and includes Chisel, an experimental 3D editor that lets users block out spatial structures before AI fills in the visual details.

World Labs is founded by AI pioneer Fei-Fei Li, who created ImageNet in 2009. The company positions spatial intelligence as the next major layer of AI development, essential for AI systems that must operate in real-world environments.

Something to think about: This marks the emergence of a new narrative surface: entire AI-generated worlds. Soon, people in opinion shaping industries won’t just build ads, scripts, or videos. They’ll shape environments. Imagine explorable political simulations (”walk through the impact of this regulation” or “attend a virtual roundtable for candidate Smith during the primary”), policy education worlds, stakeholder journey experiences, or immersive advocacy storytelling. And yes, once these worlds become persistent, there will be ad real estate everywhere: virtual billboards, product placement, sponsored sequences, and integrated narrative touchpoints. We are moving from influencing feeds to influencing world models.

Read more: World Labs Marble Announcement

3. B2B Marketers Use AI as a “Creative Sparring Partner”

Marketing Week reports that B2B teams now use AI as a structured ideation engine, rapidly generating message variants, emotional angles, and creative frames long before final copywriting.

According to Marketing Week’s State of B2B Marketing survey, 79.8% of B2B brands use AI for content writing, while 70.1% use it for developing creative ideas. Mark Barry, senior vice-president of sales and managing director of EMEA at HubSpot, notes that “marketers are using AI to brainstorm campaign themes, test different messaging and analyze what content resonates best with their audiences.”

However, 15.2% of B2B marketers have reduced agency spend due to increased AI usage, while only 1.9% have increased it. In large organizations (250 employees or more), 17% have cut agency budgets.

Something to think about: This is a good template for DC persuasion work. Run your issue frames through a “message gym”: generate 50 versions, sort by emotional resonance, stress-test them, then refine. The team that explores the most narrative branches wins. AI becomes a force multiplier for the velocity of ideation.

Read more: Marketing Week on B2B AI Usage

4. Guardian: AI-Generated Campaign Videos Are Already Normalized

The Guardian reports that AI-generated political videos are already showing up in real campaigns and are rapidly becoming normalized. What once felt fringe is becoming standard practice.

The article documents multiple campaigns using AI video tools for candidate messaging, issue advocacy, and opposition research. Platform tools have made professional-quality video generation accessible to individual candidates and small campaigns, not just well-funded operations.

Something to think about: What feels taboo to some is already very much in play. For those watching and wincing, you may have seen that AI-generated ‘Sombrero Schumer and Jeffries’ made a comeback. The campaigns that treat AI video as a strategic asset, not a guilty secret or ‘no go,’ will own the attention landscape.

Read more: The Guardian on AI Campaign Videos

5. Matthew McConaughey Licenses His Voice to ElevenLabs—Audio Becomes the Next Persuasion Battleground

Matthew McConaughey just licensed his voice to ElevenLabs, meaning we will increasingly hear “him” in media he never recorded. A Deezer/Ipsos survey across eight countries found that 97% of people cannot reliably distinguish AI-generated music from human-composed tracks, while tens of thousands of AI tracks flood streaming platforms daily.

We’re entering an era where likeness is the product, and audiences care less about how it was produced or whether something is AI-generated. Now, it’s more about whether the media is compelling.

Something to think about: Audio is now a new persuasion opportunity. You can rapidly expand share of voice using AI-generated music beds, voiceovers, and audio identity—but you must assume your principal’s voice will eventually be cloned. This unlocks enormous upside for campaigns and advocacy: sonic branding, localized variants, podcasts, infinite audio assets.

Read more: McConaughey and ElevenLabs | Reuters on AI Music Detection

6. DIRECTV Launches “Elect”—AI-Powered CTV Targeting for Political Ads

DIRECTV launched Elect, an AI-driven CTV advertising product that blends LLM insights, voter data, social indicators, and behavioral modeling to target political ads across its massive streaming footprint.

The platform uses AI modeling to segment audiences, test multiple creative variants, and measure lift like digital performance marketing. It represents the convergence of traditional TV reach with programmatic precision.

Something to think about: With AI modeling, you can target segments, test multiple creative variants, and measure lift like digital performance marketing. Campaigns should treat CTV as the new battleground between traditional persuasion and algorithmic micro-targeting.

Read more: DIRECTV Elect Announcement

Thanks for reading this edition of The Influence Model. I’ll share more about what I’m building as things take shape.

Best,
Ben