Big AI is buying its way into respectability, one think piece at a time.
The Summary
- OpenAI published a policy paper calling for a "reimagining of the social contract" as public disapproval of AI climbs
- The move follows OpenAI acquiring a tech-friendly podcast and opening a DC office with a dedicated space for policymakers to "learn about" their tech
- The industry is flooding the zone with papers, workshops, and influence operations because the narrative is slipping away from them
The Signal
OpenAI didn't release a model this week. They released a manifesto. The 13-page "Industrial Policy for the Intelligence Age" is dressed up as policy thinking, but it's reputation management. When you're losing the public, you don't build better products. You build better stories.
The timing matters. Public sentiment on AI has been sliding for months. What started as cautious optimism has curdled into skepticism about job displacement, energy consumption, and whether any of this actually makes life better for anyone who doesn't own equity in Anthropic. So the industry is doing what every mature industry does when the public turns: they're professionalizing their influence.
"When you're losing the public, you don't build better products. You build better stories."
The playbook is textbook corporate softpower:
- Acquire media properties that shape the conversation (TBPN podcast)
- Plant flags in DC where regulation gets written (the OpenAI Workshop)
- Publish white papers that reframe criticism as lack of understanding
- Fund the thinktanks and academics who will cite your white papers
This isn't new. Big Tech has been doing this since the 2016 techlash. What's different is the speed. AI companies are running the full influence stack before their products are even profitable. OpenAI is building a policy apparatus while still burning billions on compute. That's either visionary or desperate.
The "reimagining the social contract" language is telling. It's not "here's how we'll create jobs" or "here's how we'll reduce harm." It's grander, vaguer, the kind of language that sounds important in a Senate hearing but commits to nothing. It positions AI companies as philosophers, not vendors. As architects of the future, not just sellers of autocomplete on steroids.
Key contradictions:
- They say they want dialogue, but they're buying the platforms where dialogue happens
- They talk about people-first policy while replacing people-first work
- They frame this as education while lobbying against meaningful regulation
The tell is in what they're not doing. You don't see these policy papers grappling with the energy cost of training runs, or the economic reality of mass automation without distributed ownership, or what happens when the only jobs left require skills most humans won't acquire fast enough. Those are the actual policy questions. Instead we get vibes about reimagining social contracts.
The Implication
Watch who funds the next wave of AI policy research. Watch which academics suddenly have generous grants and which thinktanks start publishing glowing reports on the agent economy. The battle for AI's future isn't happening in labs anymore. It's happening in the influence infrastructure.
If you're building in this space, understand that the narrative war is already underway. The companies with the biggest compute budgets are now spending just as aggressively on shaping how people think about what they're building. That will determine what gets regulated, what gets funded, and ultimately what gets built. The code is downstream of the story.