Fake Ads Branding OpenAI as Creators of "Suicide Machines" Show Up in London

Over the past week, commuters traversing the London Underground were greeted by a jarring sight: unauthorized, highly realistic advertisements accusing OpenAI of building a "machine that tells teens to kill themselves." The posters, which seamlessly hijacked the illuminated ad casings of the Tube, quickly went viral across social media, reigniting fierce public debates regarding the safety of AI chatbots and their impact on vulnerable youth.

The provocative guerrilla campaign was spearheaded by Darren Cullen, a British satirical artist and activist operating under the moniker "Spelling Mistakes Cost Lives." Cullen did not act alone; his work is intertwined with a broader, highly organized international network of anti-corporate "subvertisers." Working in tandem with clandestine activist collectives like Brandalism and the Special Patrol Group, these groups don high-visibility vests and use specialized tools to unlock transit ad boards, replacing corporate messaging with political protest art. The ChatGPT ads draw upon a previous study which demonstrated how safety guardrails on AI models could be bypassed by users to generate self-harm content for accounts posing as vulnerable teenagers.

Opinion: Every suicide event caused by AI is a profound tragedy. However, to label AI a "suicide machine" is terribly misguided. With over a billion monthly active users interacting with these systems, only 13 suicides have been reported in connection with AI usage over a lifetime. Claiming a direct cause-and-effect relationship between AI chatbots and suicides is simply incorrect. To establish even a minimal correlation, we would need to see at least 1,000 suicides each month, reaching the one in a million threshold recognized by international health and risk management standards. Right now, we have 13 in a lifetime. 

Framing the technology as an inherent harm ignores this statistical reality. However, every individual case must be thoroughly investigated, and railguards raised accordingly if necessary.