"Anthropic Made a Great Ad. That's the Problem."

Anthropic just dropped a new commercial that is, by most creative standards, genuinely good. Created by Mother London, the ad opens with sharp piano notes and a cascade of worried voices — people asking hard questions about AI and trust, jobs, and what this technology means for society. Then it pivots. The questions turn hopeful: What if AI could help us connect? What if it could teach better than any human? What if it could help cure diseases? It's a polished, emotionally intelligent piece of brand work. And that's exactly why it backfired.

The harder an AI safety company tries to sell optimism, the louder its past warnings echo back.

The fundamental problem isn't the ad's execution. It's that Anthropic spent years building a brand around being the grown-ups in the room — the company that warns about catastrophic risks, that sues governments to slow things down, that positions Claude as the AI that won't sell you out to advertisers. When that same company suddenly runs a glossy feel-good spot about AI's bright future, people notice the whiplash. Critics were quick to point out a uncomfortable symmetry: the "gloom and doom" that fills the first half of the commercial is, in a very literal sense, what powers Anthropic's business model. Enterprises pay premium API rates precisely because they're afraid of what uncontrolled AI might do. The fear is the product.

This is the Catch-22 of AI safety branding. If you run ads that lean into the danger narrative, you're fear-mongering to sell API access — not a great look for a public benefit corporation. If you run ads that lean into optimism, you're suddenly downplaying the very risks your CEO has spent years testifying about before Congress. There is no comfortable middle, because the advertising format itself demands simplification, and "we're building something that might be incredibly powerful or might destroy everything, and we're not entirely sure which, but please subscribe to Claude Pro" doesn't fit in a 60-second spot.

The backlash also taps into something much bigger than one commercial. Public sentiment toward AI companies has soured considerably over the past year. Data center fights have moved from niche local zoning disputes to national political issues. The conversation around AI-driven job displacement has shifted from abstract futurism to concrete anxiety in industries like customer service, translation, and junior-level legal and engineering work. Mark Cuban recently framed this in stark economic terms, arguing that the resistance to AI isn't really about power grids or water consumption — it's about wealth concentration. People see a handful of companies and investors capturing enormous value while communities absorb the disruption, and they're angry about it.

Cuban's prescription is worth taking seriously. He has suggested that AI companies need to get out of their glass offices and do community tours — not to pitch products, but to genuinely engage with the workforce transitions their technology is accelerating. Show up in towns where call centers and paralegal offices are shrinking. Fund retraining programs that aren't just PR stunts. Treat the social contract as part of the product roadmap, not a public relations afterthought. This is a much harder ask than running a well-produced ad, but it's also the only thing that might actually move the needle on trust.

Anthropic is in a particularly tricky spot because it led with values harder than anyone. The company chose the public benefit corporation structure. Its CEO Dario Amodei has been one of the most vocal advocates for AI regulation. Claude famously stayed out of the advertising-in-search-results game that other chatbots embraced. These weren't cynical moves — they reflected a genuine institutional commitment to doing things differently. But they also set an extraordinarily high bar. When you define yourself as the ethical alternative, every decision gets scrutinized through that lens. A company like OpenAI or Google can run a generic brand ad and barely register. Anthropic runs one and gets a Fast Company essay asking whether advertising itself is the wrong answer.

There's something genuinely instructive here about the limits of communications strategy. In most industries, if you have a public perception problem, you hire a good agency and you run a good campaign and things improve. But AI might be the first technology where awareness is the problem — more people paying attention means more people confronting their fears, not fewer. The traditional PR playbook of "educate the public about benefits" runs straight into a public that feels increasingly educated about the costs. No amount of elegant cinematography changes the fact that people are watching an industry promise miracles while steadily eroding the economic ground beneath their feet.

Anthropic's ad isn't bad. If anything, it's too good at doing what ads do — smoothing over complexity, finding an emotional hook, making you feel good about the brand. But that's exactly the wrong move when your audience is hungry for candor, not polish. The most effective "ad" Anthropic could run right now might be a livestream of its safety team having an honest, uncomfortable conversation about what keeps them up at night — no piano score, no pivot to hope, just the weight of the responsibility they've taken on.

Further reading: Fortune on Mark Cuban's AI backlash warning, B&T on the Mother London campaign, and The Stable on the "Keep Thinking" brand platform.