The Wrong Lever: Why the AI Safety Movement Is Talking to the Wrong Audience
Description
The AI race is primarily driven by government decision-makers, corporate leaders, and regulatory bodies, not public opinion on platforms like TikTok. The 'plzdontkillus' theory of change—which assumes that raising public awareness and pressure will force policymakers to prioritize safety—may be misaligned because actual decisions are made by a small set of actors (e.g., White House officials, tech executives, and regulators) who are more influenced by geopolitical competition, market incentives, and internal bureaucratic dynamics. Public attitudes, while skeptical of AI, have limited direct impact on the pace and direction of AI development.
Sources & further reading (22)
- Winning the Race AMERICA’S AI ACTION PLAN JULY 2025https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf
- “Winning the Race: America’s AI Action Plan” – Key Pillars, Policy Actions, and Future Implications | Insights | Ropes & Gray LLPhttps://www.ropesgray.com/en/insights/alerts/2025/07/winning-the-race-americas-ai-action-plan-key-pillars-policy-actions-and-future-implications
- Factions inside the Trump administration wrestle over how to handle AI - POLITICOhttps://www.politico.com/news/2026/05/28/it-isnt-canceled-inside-the-white-house-divisions-on-ai-00938557
- What Americans Really Think About AI Algorithms: Public Confidence and Transparency in Government - Cornell Brooks Public Policyhttps://publicpolicy.cornell.edu/masters-blog/what-americans-really-think-about-ai-algorithms-public-confidence-and-transparency-in-government/
- Three Rulebooks, One Race: AI Regulation in the U.S., EU, and China – Communications of the ACMhttps://cacm.acm.org/news/three-rulebooks-one-race-ai-regulation-in-the-u-s-eu-and-china/
- Navigating the Shifting Regulatory Landscape: Investor Preparedness in the AI Sector Amid Policy-Driven Market Shiftshttps://www.ainvest.com/news/navigating-shifting-regulatory-landscape-investor-preparedness-ai-sector-policy-driven-market-shifts-2510/
- The AI safety crisis hiding behind trillion-dollar valuations | by Nayeem Islam | Mediumhttps://medium.com/@nomannayeem/the-ai-safety-crisis-hiding-behind-trillion-dollar-valuations-358e7fd0718e
- Key findings on public attitudes towards AIhttps://mbs.edu/faculty-and-research/trust-and-ai/key-findings-on-public-attitudes-towards-ai
- How do citizens perceive the use of Artificial Intelligence in public sector decisions? - ScienceDirecthttps://www.sciencedirect.com/science/article/abs/pii/S0740624X23001065
- AI risks, opportunities, regulation: Views of US public and AI experts | Pew Research Centerhttps://www.pewresearch.org/internet/2025/04/03/views-of-risks-opportunities-and-regulation-of-ai/
- The EU’s AI Power Play: Between Deregulation and Innovation | Carnegie Endowment for International Peacehttps://carnegieendowment.org/research/2025/05/the-eus-ai-power-play-between-deregulation-and-innovation?lang=en
- What TikTok's US Spin-Off Means for AI Regulation | AI Magazinehttps://aimagazine.com/news/what-tiktok-spin-off-means-for-ai-regulation
- PauseAI ⏸ on X: "Those who organised the AI Safety Summit say it achieved all of their goals. But we still don't have any meaningful protections against the worst risks from AI. A core problem is self-censorship. Incentives are misaligned between government officials and the public. Officials" / Xhttps://x.com/PauseAI/status/1724447106696405155
- AI safety undervalues foundershttps://www.lesswrong.com/posts/yw9B5jQazBKGLjize/ai-safety-undervalues-founders
- AI Safety Has 12 Months Lefthttps://www.lesswrong.com/posts/rWc6BT8BcoPhwcbSs/ai-safety-has-12-months-left
- AI Risks that Could Lead to Catastrophe | CAIShttps://safe.ai/ai-risk
- The Real Stakes of the AI Race | Foreign Affairshttps://www.foreignaffairs.com/united-states/real-stakes-ai-race
- White House unveils comprehensive AI strategy: “Winning the race: America’s AI action plan” | White & Case LLPhttps://www.whitecase.com/insight-alert/white-house-unveils-comprehensive-ai-strategy-winning-race-americas-ai-action-plan
- Beyond the Space Race: Collaboration and Competition in the Future of AI Governance • Stimson Centerhttps://www.stimson.org/2025/collaboration-and-competition-future-of-ai-governance/
- The empty national AI policy framework: Who is in charge of those in charge? | Brookingshttps://www.brookings.edu/articles/the-empty-national-ai-policy-framework-who-is-in-charge-of-those-in-charge/
- Theory of Change for AI Safety Camphttps://www.lesswrong.com/posts/Kd2cbLXQxCCRRQDcH/theory-of-change-for-ai-safety-camp
- Theory of Change | Convergence Analysishttps://www.convergenceanalysis.org/theory-of-change
Script
Cold open
What if the people you're trying to convince to slow down AI aren't the ones who actually decide how fast it goes?
Frame
The AI safety movement bets on public pressure, but the real decisions are made by a handful of officials and executives more worried about China. The twist: the very data we use to claim public concern may be irrelevant to the people who matter.
Who actually controls the pace of the AI race?
Who actually controls the pace of the AI race? The White House released 'Winning the Race: America’s AI Action Plan' in July 2025, outlining policy goals to win the AI race. But internal White House factions are wrestling over how to handle AI—the real decisions are made in these closed-door battles.
How much does public opinion really matter to these decision-makers?
How much does public opinion really matter to these decision-makers? 77% of Americans distrust both businesses and government agencies to use AI responsibly. People trust technical ability but are skeptical about safety and security—yet opinion polls showing growing concern are often ignored by those in power.
What are the real incentives driving the people in charge?
What are the real incentives driving the people in charge? Financial pressure pushes companies to prioritize capability over safety. U.S. tech giants influence policy and shape information, while incentives between officials and the public are misaligned. AI capabilities outpace safety measures, and corporations face incentives to automate labor, risking mass unemployment.
Why can't the public's safety concerns translate into policy action?
What does it actually look like when AI safety advocacy does reach the people who matter? It looks like researchers engaging directly with EU regulators during the AI Act drafting process. It looks like OpenAI's charter structuring governance constraints — constraints that even its own board eventually tried to override. It looks like the departure of Jan Leike from OpenAI in 2024, who said safety 'took a back seat to shiny products' — a defection that landed inside the labs and among funders in a way that no social media campaign has. The interventions that moved the needle targeted the handful of people actually in the room.
Turn
Instead of trying to mobilize public opinion to pressure decision-makers, a more effective lever may be to directly influence the incentive structures of the small number of actors who control the AI race. Imagine a 'safety dividend' that rewards companies for demonstrable safety practices with faster regulatory approvals or preferential access to government contracts. That aligns profit motives with safety outcomes — and it doesn't require convincing a mass audience first.
Closer
The people who decide how fast this goes aren't waiting to be convinced by a viral video. They're waiting for a competitive landscape that makes slowing down feel survivable. That is a different problem, and it probably has a different solution.