Anthropic’s newest ad is creeping people out

The Self-Awareness Trap

Anthropic has built its entire brand on being the adult in the room. While OpenAI chases Hollywood deals and Google scrambles to fix its search summaries, Anthropic usually sits in the corner talking about Constitutional AI and safety layers. They want you to think of them as the responsible technicians, not the hype-men. But their latest advertising campaign suggests they are trying a dangerous new tactic: leaning into the creepiness of their own product.

Marketing in the AI space is currently a mess of generic futuristic vistas and soft-focus lifestyle shots. Anthropic is breaking that mold by acknowledging that people are, quite frankly, weirded out by large language models. The problem is that by trying to be meta-aware of how unsettling AI can be, they are inadvertently highlighting the gap between what these models are and what we want them to be. For founders, there is a lesson here about the limits of irony in a nascent industry.

The Pivot from Safety to Edgy

For the last couple of years, Claude has been positioned as the empathetic, steerable alternative to GPT. The marketing reflected that—clean, professional, and almost clinical. This new direction is a departure. By using discomfort as a hook, Anthropic is attempting to pull off a psychological trick. They want to say, We know this is strange, and because we know it, you can trust us to build it.

This is a pivot away from technical safety and toward brand vibe. As a builder, this shift should catch your attention. When a company stops talking about the performance of the model and starts talking about the social perception of the model, it usually means they are reaching a plateau in user growth. They are trying to capture the segment of the population that is skeptical of AI by validating their fears. It is an honest approach, but it is also a desperate one.

Why Builders Should Care

If you are building on top of the Claude API, this marketing shift matters because it dictates the narrative of the tools you are selling to your own customers. If the primary provider of the intelligence is leaning into the creepy factor, you have to work twice as hard to convince your end-users that your specific implementation is safe and reliable. Anthropic is effectively poisoning their own well to look cool to a cynical audience.

• Brand identity vs. Utility: Successful tools usually don't need to apologize for existing. Think about early cloud computing or the first mobile apps. They focused on what they could do, not how people felt about them.
• The Uncanny Valley: Anthropic's new ads dwell in the uncanny valley. For a company that prides itself on human-centric AI, leaning into the robotic, slightly unsettling nature of these systems feels counter-intuitive.
• Market Saturation: This move suggests that the race for the most capable model is secondary to the race for the most recognizable brand. That is a dangerous sign for technical progress.

The Myth of the Ethical Foil

Anthropic has always framed itself as the ethical foil to the rest of Silicon Valley. This campaign is a culmination of that strategy. They are trying to position themselves as the only company that gets it. The logic goes like this: if we are honest about how weird this is, we must be the honest guys in the industry. It is a clever bit of PR, but it doesn't actually change the underlying technology.

From a founder’s perspective, I am skeptical. You cannot market your way out of the fundamental limitations of an LLM. Whether the ad is comforting or creepy, the model still hallucinates. It still has high latency. It still costs a lot to run. These are the problems that matter to builders. The marketing fluff is just a distraction from the fact that the engineering gains are becoming incremental rather than exponential.

Trust is a Function of Performance

Real trust in software isn't built through clever ads or self-referential humor. It is built through uptime, accuracy, and predictability. Anthropic’s new campaign feels like it was designed by a creative agency that doesn't spend much time in a terminal. They are selling a vibe, not a solution. For those of us writing code and building products, a vibe is useless.

The moment a technology company starts prioritizing its public image over its technical reliability is the moment it becomes a legacy company.

We saw this with the early web. The companies that tried to lead with heavy branding and emotional appeals were usually the ones that didn't have the best infrastructure. The companies that won—the ones still here today—were the ones that focused on making the utility so high that the brand became secondary. Google didn't need to tell us that search was weird or spooky; they just made it work.

The Takeaway for Founders

The lesson here for builders is simple: do not mistake brand awareness for product-market fit. Anthropic is feeling the pressure to stand out in a world where every LLM is starting to look and act the same. Their response is to get weird. Your response should be to stay focused on the user’s pain points.

If you are building an AI-native startup right now, ignore the ad campaigns. Do not try to be the edgy, self-aware version of your competitors. Just be the version that works. While the big labs spend millions of dollars on ads that creep people out, you can win by providing the boring, stable, and incredibly useful services that people actually need. In the end, utility wins over irony every single time.

Read the original at TechCrunch AI →