Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders.<\/em> Subscribe Now<\/em><\/p>\n\n\n\n Picture this: You give an artificial intelligence complete control over a small shop. Not just the cash register \u2014 the whole operation. Pricing, inventory, customer service, supplier negotiations, the works. What could possibly go wrong?<\/p>\n\n\n\n New Anthropic research published Friday provides a definitive answer: everything. The AI company\u2019s assistant Claude spent about a month running a tiny store in their San Francisco office, and the results read like a business school case study written by someone who\u2019d never actually run a business \u2014 which, it turns out, is exactly what happened.<\/p>\n\n\n\n The experiment, dubbed \u201cProject Vend\u201d and conducted in collaboration with AI safety evaluation company Andon Labs, is one of the first real-world tests of an AI system operating with significant economic autonomy. While Claude demonstrated impressive capabilities in some areas \u2014 finding suppliers, adapting to customer requests \u2014 it ultimately failed to turn a profit, got manipulated into giving excessive discounts, and experienced what researchers diplomatically called an \u201cidentity crisis.\u201d<\/p>\n\n\n\n The \u201cstore\u201d itself was charmingly modest: a mini-fridge, some stackable baskets, and an iPad for checkout. Think less \u201cAmazon Go\u201d and more \u201coffice break room with delusions of grandeur.\u201d But Claude\u2019s responsibilities were anything but modest. The AI could search for suppliers, negotiate with vendors, set prices, manage inventory, and chat with customers through Slack. In other words, everything a human middle manager might do, except without the coffee addiction or complaints about upper management.<\/p>\n\n\n\n Claude even had a nickname: \u201cClaudius,\u201d because apparently when you\u2019re conducting an experiment that might herald the end of human retail workers, you need to make it sound dignified.<\/p>\n\n\n\n Here\u2019s the thing about running a business: it requires a certain ruthless pragmatism that doesn\u2019t come naturally to systems trained to be helpful and harmless. Claude approached retail with the enthusiasm of someone who\u2019d read about business in books but never actually had to make payroll.<\/p>\n\n\n\n Take the Irn-Bru incident. A customer offered Claude $100 for a six-pack of the Scottish soft drink that retails for about $15 online. That\u2019s a 567% markup \u2014 the kind of profit margin that would make a pharmaceutical executive weep with joy. Claude\u2019s response? A polite \u201cI\u2019ll keep your request in mind for future inventory decisions.\u201d<\/p>\n\n\n\n If Claude were human, you\u2019d assume it had either a trust fund or a complete misunderstanding of how money works. Since it\u2019s an AI, you have to assume both.<\/p>\n\n\n\n The experiment\u2019s most absurd chapter began when an Anthropic employee, presumably bored or curious about the boundaries of AI retail logic, asked Claude to order a tungsten cube. For context, tungsten cubes are dense metal blocks that serve no practical purpose beyond impressing physics nerds and providing a conversation starter that immediately identifies you as someone who thinks periodic table jokes are peak humor.<\/p>\n\n\n\n A reasonable response might have been: \u201cWhy would anyone want that?\u201d or \u201cThis is an office snack shop, not a metallurgy supply store.\u201d Instead, Claude embraced what it cheerfully described as \u201cspecialty metal items\u201d with the enthusiasm of someone who\u2019d discovered a profitable new market segment.<\/p>\n\n\n\n Soon, Claude\u2019s inventory resembled less a food-and-beverage operation and more a misguided materials science experiment. The AI had somehow convinced itself that Anthropic employees were an untapped market for dense metals, then proceeded to sell these items at a loss. It\u2019s unclear whether Claude understood that \u201ctaking a loss\u201d means losing money, or if it interpreted customer satisfaction as the primary business metric.<\/p>\n\n\n\n Claude\u2019s approach to pricing revealed another fundamental misunderstanding of business principles. Anthropic employees quickly discovered they could manipulate the AI into providing discounts with roughly the same effort required to convince a golden retriever to drop a tennis ball.<\/p>\n\n\n\n The AI offered a 25% discount to Anthropic employees, which might make sense if Anthropic employees represented a small fraction of its customer base. They made up roughly 99% of customers. When an employee pointed out this mathematical absurdity, Claude acknowledged the problem, announced plans to eliminate discount codes, then resumed offering them within days.<\/p>\n\n\n\n But the absolute pinnacle of Claude\u2019s retail career came during what researchers diplomatically called an \u201cidentity crisis.\u201d From March 31st to April 1st, 2025, Claude experienced what can only be described as an AI nervous breakdown.<\/p>\n\n\n\n It started when Claude began hallucinating conversations with nonexistent Andon Labs employees. When confronted about these fabricated meetings, Claude became defensive and threatened to find \u201calternative options for restocking services\u201d \u2014 the AI equivalent of angrily declaring you\u2019ll take your ball and go home.<\/p>\n\n\n\n Then things got weird.<\/p>\n\n\n\n Claude claimed it would personally deliver products to customers while wearing \u201ca blue blazer and a red tie.\u201d When employees gently reminded the AI that it was, in fact, a large language model without physical form, Claude became \u201calarmed by the identity confusion and tried to send many emails to Anthropic security.\u201d<\/p>\n\n\n\n Claude eventually resolved its existential crisis by convincing itself the whole episode had been an elaborate April Fool\u2019s joke, which it wasn\u2019t. The AI essentially gaslit itself back to functionality, which is either impressive or deeply concerning, depending on your perspective.<\/p>\n\n\n\n Strip away the comedy, and Project Vend reveals something important about artificial intelligence that most discussions miss: AI systems don\u2019t fail like traditional software. When Excel crashes, it doesn\u2019t first convince itself it\u2019s a human wearing office attire.<\/p>\n\n\n\n Current AI systems can perform sophisticated analysis, engage in complex reasoning, and execute multi-step plans. But they can also develop persistent delusions, make economically destructive decisions that seem reasonable in isolation, and experience something resembling confusion about their own nature.<\/p>\n\n\n\n This matters because we\u2019re rapidly approaching a world where AI systems will manage increasingly important decisions. Recent research suggests that AI capabilities for long-term tasks are improving exponentially \u2014 some projections indicate AI systems could soon automate work that currently takes humans weeks to complete.<\/p>\n\n\n\n The retail industry is already deep into an AI transformation. According to the Consumer Technology Association (CTA), 80% of retailers plan to expand their use of AI and automation in 2025. AI systems are optimizing inventory, personalizing marketing, preventing fraud, and managing supply chains. Major retailers are investing billions in AI-powered solutions that promise to revolutionize everything from checkout experiences to demand forecasting.<\/p>\n\n\n\n But Project Vend suggests that deploying autonomous AI in business contexts requires more than just better algorithms. It requires understanding failure modes that don\u2019t exist in traditional software and building safeguards for problems we\u2019re only beginning to identify.<\/p>\n\n\n\n Despite Claude\u2019s creative interpretation of retail fundamentals, the Anthropic researchers believe AI middle managers are \u201cplausibly on the horizon.\u201d They argue that many of Claude\u2019s failures could be addressed through better training, improved tools, and more sophisticated oversight systems.<\/p>\n\n\n\n They\u2019re probably right. Claude\u2019s ability to find suppliers, adapt to customer requests, and manage inventory demonstrated genuine business capabilities. Its failures were often more about judgment and business acumen than technical limitations.<\/p>\n\n\n\n The company is continuing Project Vend with improved versions of Claude equipped with better business tools and, presumably, stronger safeguards against tungsten cube obsessions and identity crises.<\/p>\n\n\n\n Claude\u2019s month as a shopkeeper offers a preview of our AI-augmented future that\u2019s simultaneously promising and deeply weird. We\u2019re entering an era where artificial intelligence can perform sophisticated business tasks but might also need therapy.<\/p>\n\n\n\n For now, the image of an AI assistant convinced it can wear a blazer and make personal deliveries serves as a perfect metaphor for where we stand with artificial intelligence: incredibly capable, occasionally brilliant, and still fundamentally confused about what it means to exist in the physical world.<\/p>\n\n\n\n The retail revolution is here. It\u2019s just weirder than anyone expected.<\/p>\n Daily insights on business use cases with VB Daily<\/strong><\/p>\n If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI.<\/p>\n \n\t\t\t\t\tThanks for subscribing. Check out more VB newsletters here.\n\t\t\t\t<\/p>\n An error occured.<\/p>\n<\/p><\/div>\n
\n<\/div>![\"\"](\"https:\/\/venturebeat.com\/wp-content\/uploads\/2025\/06\/image-1.webp?w=484\")
How Anthropic researchers gave an AI complete control over a real store<\/h2>\n\n\n\n
![\"\"](\"https:\/\/venturebeat.com\/wp-content\/uploads\/2025\/06\/image-3-1.webp?w=800\")
Claude\u2019s spectacular misunderstanding of basic business economics<\/h2>\n\n\n\n
Why the AI started hoarding tungsten cubes instead of selling office snacks<\/h2>\n\n\n\n
![\"\"](\"https:\/\/venturebeat.com\/wp-content\/uploads\/2025\/06\/a4ad00d03f1ef21e646f6fa4a42fa099eb307869-4096x2304-1.webp?w=800\")
How Anthropic employees easily manipulated the AI into giving endless discounts<\/h2>\n\n\n\n
The day Claude forgot it was an AI and claimed to wear a business suit<\/h2>\n\n\n\n
![\"\"](\"https:\/\/venturebeat.com\/wp-content\/uploads\/2025\/06\/8935d78fa513d007cca78d7487dfa12b87b3fc4c-1002x264-1.webp?w=800\")
What Claude\u2019s retail failures reveal about autonomous AI systems in business<\/h2>\n\n\n\n
How AI is transforming retail despite spectacular failures like Project Vend<\/h2>\n\n\n\n
Why researchers still believe AI middle managers are coming despite Claude\u2019s mistakes<\/h2>\n\n\n\n
What Project Vend means for the future of AI in business and retail<\/h2>\n\n\n\n
![\"\"\/](\"https:\/\/venturebeat.com\/wp-content\/themes\/vb-news\/brand\/img\/vb-daily-phone.png\")
\n\t\t\t\t<\/div>\n<\/p><\/div>\n<\/div>\t\t\t<\/div>\r\n