What happens when a decades-long corporate dispute collides with generative artificial intelligence? The evolving confrontation between activist John Donovan and Shell has become an unusual but revealing test case. What began in the 1990s as a traditional legal and reputational conflict has, in recent years, migrated into the domain of AI systems—where archives, allegations, and corporate responses are reinterpreted, recombined, and sometimes distorted by large language models.
Central to this latest phase is what participants have termed a “bot war”: a series of experiments in which Donovan inputs historical materials, leaked documents, and prior claims into multiple AI platforms, then publishes the outputs—highlighting inconsistencies, apparent hallucinations, and divergent narratives about Shell’s past conduct. The results raise broader questions about how AI systems handle contested histories, how corporate reputations can be affected by machine-generated content, and where responsibility lies when outputs blur the line between synthesis and fabrication.
Screenshot from the 
