When the word 'safety' backfires: what the airline industry's lesson means for Anthropic

Airlines learned long ago never to use the word 'safety' in their advertising — the word itself conjures up in a passenger's mind the very image it's trying to avoid: a crash. According to historian Richard Popp's research on airline advertising, by the 1930s airlines followed an unwritten rule of never showing mountains, night flights, large bodies of water, or maintenance work in their ads — because each of these, even without anyone uttering the word 'safety,' could summon the thought of disaster.

Despite that taboo, Pan Am ran an ad campaign in the 1980s touting its own safety amid a wave of hijackings and bomb threats targeting transatlantic flights. Then, in December 1988, the bomb that brought down Pan Am Flight 103 over Lockerbie, Scotland, killed 270 people. Aviation safety expert Glen Winning argues it was that tragedy that convinced the entire industry to abandon the word 'safe' altogether.

A similar dilemma is now playing out in artificial intelligence. Anthropic co-founder and CEO Dario Amodei has consistently positioned the company as the AI industry's most safety-committed player — he even left OpenAI in 2021 specifically over disagreements about its approach to safety. But that strategy may now be working against him.

A recent Gallup poll found that nearly half of Americans under 30 now believe AI does more harm than good — a sharp jump from a year earlier. Pew Research data from 2026 is even more troubling: most Americans believe AI is advancing too fast. A Quinnipiac poll found that more than three-quarters of Americans say they trust AI only 'rarely' or 'sometimes' — even as actual usage keeps climbing steadily.

PR firms have long known what's called the 'association problem': talking about a risk in connection with yourself — even to deny it — plants that very risk firmly in the listener's mind. As linguist George Lakoff wrote in his 2004 book 'Don't Think of an Elephant,' telling someone not to think of an elephant conjures up, above all, an elephant — and repeating the instruction doesn't erase the image, it reinforces it. Anthropic's TV ad, 'There Is Hope in Hard Questions,' opens on a burning house and moves through a crowd being scanned by facial recognition, a person sleeping at a bus stop, and a cemetery-like landscape — all under the question, 'Can AI be trusted?' OpenAI CEO Sam Altman has said that when he first saw the ad, he thought it was satire.

Airlines eventually realized that talking about a risk in connection with yourself can do more damage than saying nothing at all — because the word itself does the very thing it's trying to avoid. The AI industry, it seems, may have to relearn that same lesson.