Automating the Absurd: How SSE Let a PhD Student and a Bot Win in Court
AI-generated image · US National Wire
A British energy giant spent three years chasing a non-existent address, proving that AI isn't the problem—corporate incompetence is.
In the crypto and web3 world, we talk a lot about 'hallucinations' as the primary failure point of AI. But as a realist who has seen enough corporate theater to last a lifetime, I can tell you that the real danger isn't the AI making things up—it's the corporate machine automating stupidity at a scale that defies logic.
Take the case of SSE Energy Supply. As The Register first reported, this energy company spent nearly three years pursuing a debt from a non-existent address. The company targeted Lyle Hopkins, who is pursuing a doctorate at the University of Oxford. The source of the conflict? An unused electricity meter associated with a previous owner's failed attempt to convert a property into flats. Specifically, SSE insisted there was a 'unit 8b' at the property; Hopkins maintained there was no such thing.
For over 20 months, SSE billed Hopkins at business rates, totaling £1,091.01. The sheer stubbornness of the corporate apparatus is staggering: The Register reports that an SSE representative even sent an email on June 15, 2024, explicitly stating that Hopkins was not liable. Despite this internal admission, the company proceeded to deploy debt collectors to recover funds for a disconnected meter at a location that didn't exist.
***
**Nate's Take: The Human Error Multiplier**
*Opinion: This isn't a story about AI 'disrupting' the law; it's a story about a company being so fundamentally broken that a single student with a few API subscriptions could dismantle their legal standing. AI didn't create the hallucination here—SSE's billing department did. The AI was simply the tool used to audit the wreckage.*
***
Facing harassment and unable to afford a solicitor, Hopkins—a software engineer and PhD student at Oxford Interdisciplinary Bioscience DTP—turned to the tools of the trade. He utilized GPT-5.5 and Claude Fable to navigate the legal justifications and court rules. Hopkins told The Register that while early GPT models struggled with case law, he used the AI to test SSE's claims against its own records and to find the necessary legal justifications.
He didn't just blindly trust the bot. Hopkins emphasized that verification was the key, noting he had to steer the AI and ensure the cited case law actually existed. The cost of this 'legal team' was roughly £175 in API fees and subscriptions to OpenAI, Anthropic, and GitHub Copilot.
On July 17, a judge at the County Court at Oxford (St Aldate) ruled in Hopkins' favor, awarding him £1,087.88, including interest and expenses. The judge described the experience as a 'rollercoaster ride' and warned that continued pursuit of the debt would constitute harassment.
In a final display of corporate dysfunction, The Register reports that SSE sent another bill with a payment demand on July 23—just six days after the court ruling.
Hopkins noted that SSE had offered him a higher settlement prior to the hearing, provided he signed confidentiality and non-disparagement clauses. He refused, choosing instead to highlight the company's failures. The toll was high: Hopkins says the ordeal cost him hundreds of hours and contributed to his medical-research doctoral funding ending before he could finish his degree. He has since launched a funding campaign to complete his doctorate.