Key takeaways
- McKinsey estimates generative AI could unlock between $240 billion and $390 billion in annual value for retail, making it one of the highest-opportunity sectors for AI adoption globally.
- The same infrastructure driving that value also builds detailed consumer profiles that, in some cases, are monetized and sold without the consumer’s explicit knowledge.
- Retail’s AI risks are not comparable to extinction-level scenarios, but they raise real questions about consent and data governance that the industry has not yet resolved.
- The governance frameworks emerging from the AI safety debate — from the EU AI Act to US congressional investigations — will govern commercial retail AI; the timeline is already running.
On September 9, 2026, Jacob Coxon, a safety researcher at Anthropic, resigned and posted a warning to colleagues. Without more caution, he wrote, superintelligent AI creates “a risk of causing human extinction.” Evan Hubinger, Anthropic’s alignment science lead, responded publicly, placing his personal estimate of that risk above ten percent over the next decade. More than 1,300 employees across frontier AI labs had already signed an open letter calling for a slowdown. The US Senate is demanding classified briefings. Legislation is moving. The White House has dismissed the warnings as the work of “negative forces.”
While that debate plays out in Washington, AI in retail is already operational. Not theoretical, but embedded across pricing, inventory, personalization, and consumer profiling at industry scale. Bipartisan lawmakers are now asking whether AI could become dangerous. In the meantime, retail has been living with that question for some time.
The commercial case for AI in retail
By most commercial measures, AI in retail has been treated as a pure opportunity. McKinsey estimates that the technology could unlock between $240 billion and $390 billion in annual value for the retail sector — the equivalent to a margin increase of 1.2 to 1.9 percentage points across the industry. Deloitte’s 2026 survey of 200 retail executives found that 82% plan to increase AI investment in the next twelve months, with sector spending projected to reach $19.9 billion this year, up from $6.4 billion in 2021.
The logic is sound. AI improves demand forecasting, reduces inventory waste, powers personalization, and frees up staff time. Adobe Analytics, tracking more than one trillion visits to US retail sites, found that during the 2025 holiday season AI-referred traffic converted 31% more frequently than other sources, and by the first quarter of 2026, traffic from AI sources to retail sites had grown 393% year over year.
At the RLC Fashion Summit in Milan last June, Neeraj Teckchandani, CEO of Apparel Group, made clear how far things have already moved: “We’ve built a state-of-the-art automated distribution center with no human involvement at all, and we’re doing full value-chain optimization: demand forecasting, pricing, location, replenishment, transfers, markdowns, everything.”
What does AI really know about you?
The data retail AI collects serves many legitimate purposes. As that data becomes more detailed and more valuable, questions of AI data security also become harder to separate from the wider privacy debate. Purchase history, browsing patterns, and location data power the personalized recommendations, tailored offers, targeted marketing, and improved customer service experiences that most consumers have come to expect.
What is less visible is the scale of what is collected. The Federal Trade Commission’s preliminary research findings published in January 2025, documented the range of consumer data that AI-powered intermediary firms — companies retailers hire to manage pricing and marketing — work with across at least 250 retail clients. The list goes well beyond purchase history: real-time location, demographic profile, annual income, relationship status, IP address, browsing history, and cursor movements on a webpage.
That data does not always stay within the retailer-customer relationship. Some retail media networks monetize consumer behavioral profiles by selling targeted access to brands and advertisers. The consumer whose data underpins that network sees none of the commercial return and, in most cases, was never explicitly told it was happening.
This is where the legitimate value exchange becomes harder to read clearly. Most consumers understand that sharing data with a retailer improves their experience. Fewer understand that the same data may be packaged, sold, and used to influence what they see across entirely different platforms. The gap between what people think they are agreeing to and what they are consenting to is where the more serious questions about AI in retail begin.
Are retail’s AI risks a different kind of danger?
Washington is debating scenarios where AI operates beyond human control: weapons development, systemic failures, the loss of oversight at a global scale. Retail’s relationship with AI raises different questions entirely: about consent, about the terms of the value exchange between consumer and retailer, and about what governance should look like when the technology is already embedded in everyday commerce.
However, asking an AI to design a weapon and asking it to personalize a marketing campaign are categorically different acts. The risk profile is different, the intent is different, and the consequences are different. Retail’s AI challenges do not belong in the same sentence as extinction risk. But that does not mean they are neutral, or that the industry has resolved them.
Emanuela Prandelli, LVMH Professor of Fashion and Luxury Management at Bocconi University, captured the operating philosophy most retailers have adopted: “The mantra today is to automate, thanks to technology, the ordinary — mainly what’s behind the scenes — and humanize the extraordinary, what has to remain rooted in emotions and experiences.”
It is a framework most of the industry operates by without much debate. The question it leaves open is how much transparency consumers are owed about what happens behind the scenes, and whether the current answer is sufficient.
Consumer research reflects a genuine unease, though not always a consistent one. AI and privacy sit at the very center of that unease. According to Alchemer’s 2026 Retail Report, a survey of 1,002 US consumers, privacy is the leading AI concern across every age group, cited by 35.1% of respondents overall and rising to 38% among 45 to 60 year-olds. Retail ranked 12th of 17 sectors in privacy maturity in TrustArc’s 2025 Global Privacy Benchmarks Survey, scoring 54% against a global benchmark of 61%. The industry’s governance has not kept pace with its adoption.
But the same survey found that 74% of 18 to 29 year-olds have already used AI to research a purchase. Consumers are using these tools daily to write, to search, to shop, while simultaneously expressing concern about what AI does with their data. The consumer who worries about AI knowing their shopping habits is often the same person who asked an AI tool to draft their last email. That is exactly what the early stages of a technology shift look like: adoption moves faster than understanding.
Why the AI debate might become retail’s problem
The most direct answer is regulatory. The governance frameworks being built in response to the AI safety conversation, designed primarily with systemic and existential risk in mind, will eventually govern commercial retail AI as well. The EU AI Act, now in active enforcement, covers retailers using AI-driven personalization, automated checkout decisioning, and customer-facing AI systems. In the US, the FTC, Department of Justice, and multiple congressional committees have opened retail-adjacent AI investigations within the past twelve months. The legislation being written does not distinguish between a weapons prompt and a recommendation engine. Instead, it governs the infrastructure.
Retail, as with most industries, built its AI infrastructure way before the regulatory frameworks were designed. Those frameworks are now here. Whether the industry views the AI safety debate in Washington and elsewhere as its concern or not, the compliance obligations emerging from it will require a response from any business running AI in retail at commercial scale.
