Anthropic's graveyard ad proves even AI safety leaders can fumble the pitch

Anthropic just learned a hard lesson about the gap between ethical branding and just creeping people out. The company’s latest ad campaign, ‘There’s hope in hard questions,’ was supposed to position it as the responsible adult in the AI room. Instead, it triggered a wave of backlash for imagery that felt less like a thoughtful meditation on risk and more like a trailer for a dystopian thriller nobody asked to see.

The spot opens on a burning house and doesn’t get cheerier from there. Viewers are served a montage of facial recognition surveillance, homelessness, rows of tombstones, and what looks like cobalt miners digging up smartphone ingredients. Over this, disembodied voices ask cheerful questions like ‘Can AI be trusted?’ and ‘Who’s gonna hit the brakes if we need to?’ The intended message — that Anthropic takes these fears seriously — got lost in the execution. One shot, appearing to show Arlington National Cemetery, proved especially radioactive. ‘I can’t stress enough how fucked up it is that Anthropic is running an ad that includes this image,’ one commenter wrote, a sentiment that echoed widely across social media.

Sam Altman wasted no time twisting the knife. The OpenAI CEO posted to X that he ‘thought this was satire’ and kept looking for a parody account handle. For a company that has built its entire identity on being the ethical alternative to rivals like OpenAI, the mockery from its chief competitor stings. The irony here is thick. Anthropic was following a playbook as old as corporate PR: name the harms your industry causes, then pitch yourself as the only one responsible enough to handle them. Oil companies do it. Tech companies do it. Usually, it’s boring but safe. This time, the tone was so self-serious and the imagery so morbid that it made the company look dangerously out of touch rather than admirably cautious.

The episode is particularly awkward given Anthropic’s recent marketing wins. During the Super Bowl, the company ran slick, funny ads taking shots at OpenAI for putting ads in ChatGPT. Those landed well. They felt human. This latest effort — which one critic suggested came from effective altruists ‘living in a bubble of ai psychosis’ — feels like it was focus-grouped in a bunker. The campaign inadvertently reinforces a suspicion that’s been dogging the AI safety crowd: that its warnings about existential risk can sound less like prudence and more like a kind of apocalyptic branding exercise. When your advertisement reminds people of the brainwashing propaganda sequence from The Parallax View, you’ve probably missed the mark on projecting benevolent intent.