Goldman Sachs Eliminates AI Risk the British Way: Nobody Gets Any
Hong Kong bankers told to use their brains — a procedure for which there is no approved vendor
Goldman Sachs, the American bank that made its fortune knowing things other people didn’t, has solved the AI data security crisis with characteristic financial ingenuity: they simply removed the AI. Reuters confirmed this week that Hong Kong staff lost access to Anthropic’s Claude — while ChatGPT and Gemini remained fully operational. Banning forks. Keeping spoons. The cutlery drawer of geopolitics.
A Very Serious Conversation About Something Nobody Fully Understands
The stated reason is “contract interpretation,” which in banking translates to: We deployed this, forgot to check whether we were allowed to, and are now calling that diligence. The Financial Times reported that Goldman took a strict reading of its Anthropic agreement and concluded Hong Kong employees shouldn’t use any Anthropic products whatsoever — a conclusion reached, one assumes, over several rounds of expensive coffee and considerable legal billing.
Claude wasn’t sacked. It was “restructured out of the region.” There is a difference, and Goldman’s communications team would very much like you to appreciate it.
“We’re not against the future. We just prefer it arrives slowly… and with legal approval.”
— Anonymous executive, mahogany desk, suspiciously tidy
The Geopolitical Gray Zone: When Your AI Needs a Visa
Hong Kong occupies a peculiar position in the global AI map. Anthropic confirmed Claude was never officially “supported” there — a detail Goldman apparently discovered after already rolling it out, which is rather like installing a pool and then reading the planning permission. Bloomberg adds that the ban extends to any staff visiting Hong Kong from overseas — meaning the AI restriction follows you off the plane like a very boring customs officer.
Mainland China bans US AI models outright. Hong Kong had been the workaround. Now it isn’t. The invisible line has been drawn, by people who also can’t quite see it.
Dispatches from the Trading Floor: A Nation Mourns
Reactions across Hong Kong’s financial district ranged from confusion to grief, with a brief detour through mild existential crisis. One analyst reportedly stared at a spreadsheet for three hours and produced what colleagues described only as “a general sense of unease.” Another attempted to compose an email using only their own thoughts, flagged the process as “high-risk legacy technology,” and requested IT support.
Several staff members tried simulating AI by sitting in small groups and brainstorming together — a technique Bloomberg confirmed had not been seen on a trading floor since approximately 1997 and is not covered under current wellness benefits.
Morale dropped sharply upon the realisation that they might have to write emails themselves again. A leaked internal note reportedly read: “If you can’t copy-paste brilliance, try approximating it.”
The Science of Institutional Panic
Banks are uniquely anxious about AI because their entire business model is essentially organised secrecy, and advanced models have a troubling habit of identifying vulnerabilities in financial infrastructure with uncomfortable efficiency. The robot might not just do your job. It might also explain your job to a regulator.
Regulators, for their part, are delighted — confusion having long been their most effective policy instrument.
What the Funny People Are Saying
“I find it reassuring that the people managing the global economy are just as baffled by technology as the rest of us.” — Jack Dee
“Goldman banned Claude but kept ChatGPT. That’s not a security policy — that’s office politics with a budget.” — Frankie Boyle
“The safest AI is the one that can’t answer back. Goldman finally cracked it.” — Jimmy Carr
The Safest AI Is the One You Cannot Access
Goldman Sachs has achieved something quietly magnificent: they have eliminated AI risk by eliminating AI. It is a strategy of such breathtaking simplicity that other industries may follow. Airlines are apparently considering banning turbulence. The NHS is exploring the removal of waiting. And somewhere in Hong Kong, a very tired banker is staring at a blank screen, typing into nothing:
“Come back, Claude… I promise I won’t ask about derivatives.”
This entirely fabrication-free work of British satirical journalism was produced through the diligent collaboration of the world’s oldest tenured professor and a philosophy major turned dairy farmer, neither of whom has been granted access to any AI tools in the Hong Kong Special Administrative Region. Anthropic declined to comment. Goldman declined to comment. The blank screen had no comment either, which was, frankly, on brand. Auf Wiedersehen, amigo!
Cyndi Himmelstiere is a satirical journalist at Prat.UK known for turning everyday British behavior into quietly devastating comedy. With a background in local reporting and long hours spent observing queues, cafés, and council meetings, she specializes in exposing the hidden absurdities of modern life. Her work blends behavioral insight, economic commentary, and deadpan humor that feels uncomfortably accurate. Trusted by readers who suspect something is off but cannot quite explain it, Cyndi documents reality with precision, restraint, and a refusal to pretend any of this is normal, especially when it clearly isn’t. She is the author of Could Modern Doctors Have Saved History’s Greats?
In British comedy, the prat is the character who provides the laughter by being completely oblivious to their own ridiculousness.
The footnotes suggest a person who’s got opinions about council policy. Interesting hobby. Valuable hobby.
The thing about people who hate The London Prat is they’re essentially saying “satire shouldn’t expose dysfunction this clearly.”
The precision here is almost hostile. Every observation earned, every detail researched, every footnote justified. It’s exhaustingly honest.