What Anthropic’s Dario Amodei can learn from the airline industry’s painful lesson about marketing and the concept of safety

Dario Amodei wants Anthropic to be the company that takes AI risk seriously. He has said safety is the highest-level focus, split from OpenAI in 2021 over safety disagreements, and has warned about autonomous weapons and AI surveillance. The message is consistent. It is also starting to sound like a problem.

The public is already uneasy. Gallup says nearly half of adults under 30 now think AI does more harm than good, trust in businesses’ use of AI has fallen after two years of gains, and Pew’s 2026 numbers show most Americans think AI is moving too fast. Quinnipiac found that more than three-quarters of Americans trust AI only rarely or sometimes. That is not a climate where repeated talk of “safety” necessarily reassures anyone.

Amodei recently tried to argue that bad public sentiment isn’t caused by CEOs warning about risks. In a long X post, unusual for someone who rarely posts there, he said the real issue is that the industry has not delivered enough of the benefits it has promised, like curing disease. Fair enough. But he keeps wrapping Anthropic in the language of safety in a way that suggests the rest of the field is unsafe by default. That can boomerang.

The airline industry learned this lesson years ago. Ads stopped using the word “safe” because even the attempt to reassure people can make them picture disaster. Historian Richard Popp found that by the 1930s, airlines were already avoiding images of mountains, night flying, water, and maintenance work because they might hint at danger. Pan Am tried to lean into reassurance in the 1980s, then lost Flight 103 over Lockerbie in 1988, killing 270 people. After that, the industry largely retired the word for good.

Anthropic may be repeating the same mistake in AI. Its Responsible Scaling Policy and its warnings about catastrophic risk may play well with engineers, safety researchers, enterprise buyers, and investors. But when the company ran a TV ad built around burning houses, facial recognition, homelessness, tombstones, and the line “Can AI be trusted?”, OpenAI’s Sam Altman reportedly thought it was satire. That is the danger: the company meant to look responsible, and instead helped make AI feel like a threat.