AI shopping agents test Amazon and Walmart’s advertising control | Magica
AI shopping agents test Amazon and Walmart’s control over retail advertising
Editorial Team
••📖7 min read
Walmart plans new permissions and a gateway for outside shopping agents, while Amazon blocks unauthorized tools and sells ads through its own assistant. The stakes are product discovery, customer data and whether delegated purchases add sales or displace advertising opportunities.
Walmart already permits outside shopping agents; its planned profiles and gateway add permission and transaction controls.
Amazon already offers advertising inside its own AI shopping assistant, showing that automated shopping can also create ad opportunities.
The unresolved question is whether outside agents bring additional sales or weaken retailers’ influence over product discovery.
Walmart’s October 21 announcement adds a permission system to an existing shopping channel: software that browses and buys on customers’ behalf. Amazon, meanwhile, has blocked Muse, Meta’s personal AI agent, from browsing and buying on its platform. Their different responses put a commercial question behind the technology: who gets to influence a purchase when the shopper delegates the work?
Company-reported Amazon advertising-services sales in Q4 2024 and Q4 2025. These totals do not measure exposure to outside shopping agents. Source: Amazon financial results.
A sale and an advertising opportunity can move apart
Both retailers combine selling goods with advertising businesses. The potential disruption is that an outside assistant could select a product before its user reaches a retailer’s own shopping interface. The retailer might still fulfill the order while losing an opportunity to influence the choice. That is a possible mechanism of pressure, rather than evidence that advertising revenue has already fallen.
The financial incentive to preserve that influence is substantial. Walmart’s fiscal 2026 results put its global advertising business at nearly $6.4 billion, up 46% for the year ended January 31, 2026. That figure includes VIZIO and is recorded partly as sales and partly as reductions to cost of sales, depending on the arrangement. It is a measure of Walmart’s broader advertising operation, not an estimate of shopping-agent exposure.
The filing also reports that Walmart Connect, its U.S. advertising business, grew 41% in the fourth quarter excluding VIZIO. That narrower measure distinguishes growth in its existing operation from the broader total that includes VIZIO.
Amazon reported $21.3 billion in advertising-services sales for the quarter ended December 31, 2025, up 23% from a year earlier, in its . These figures describe broader advertising businesses with different periods and accounting scopes; neither measures the amount vulnerable to outside shopping agents.
Five technology groups have accumulated an estimated $3.2 trillion in future purchase and lease commitments. Filings show the funding pressure, but broad spending categories, long payment schedules and a methodology change limit what the total says about AI debt and growth.
The two companies’ different access policies also have explanations beyond advertising. Amazon said it had not authorized Muse to access customer accounts, scrape data or process transactions, citing security and customer-experience concerns, according to the October account of their competing approaches. The same report quoted GlobalData Retail analyst Neil Saunders’s interpretation that Amazon likely feared damage to advertising. That is an analyst’s assessment of its incentives, rather than Amazon’s stated explanation.
Walmart’s illustration of its planned Agent Profiles shows agent purchases and customer approval controls. Image: Walmart. Source: Walmart agent access announcement.
Walmart opens a gate with conditions
Walmart, whose stores and website sell groceries, household essentials and other merchandise, had already announced an outside-AI shopping partnership a year earlier. Its October 2025 agreement with OpenAI, the company behind the ChatGPT conversational AI service, envisaged customers shopping through ChatGPT using Instant Checkout. The new initiative extends a strategy already under way. Walmart’s practical purchases and low-price positioning offer a possible advantage when customers instruct agents to find a deal, although that does not establish that agents will favor it.
On October 21, Walmart announced Agent Profiles and the Walmart Agent Gateway, intended to let customers choose agents for Walmart and Sam’s Club while controlling what those agents can do. It said profiles would launch in the following month, separating each agent’s purchases from those made by the customer. Customers could require approval for every purchase or permit spending within a limit, and revoke access at any time.
Its published terms make the limits concrete: customers must approve purchases by default, and Walmart may require additional approval even within spending limits. Agents do not receive payment-card information or permission to change payment methods, and Walmart can block agents that break its rules. The company also says its gateway will help agents find customers’ usual purchases without sharing their broader shopping history.
Walmart says agents navigating its human-facing website can be slow, make errors or get stuck. The gateway is intended to give agents a faster, more reliable connection, beginning with a beta for companies and developers. Those are planned capabilities and company claims, rather than demonstrated reliability results.
Walmart also says agents see the same product prices as people on Walmart.com, with applicable offers and fees disclosed before purchase. Its returns process applies when an agent acts unexpectedly. Information access is limited to the task and customer permission, but the agent provider’s privacy policy governs information it receives. Opening access still leaves consequential choices about pricing, returns and data with the retailer and agent provider.
Amazon’s demonstration shows a sponsored product-page prompt opening a conversation with Alexa for Shopping. Image: Amazon Ads. Source: Amazon Ads.
Amazon can sell ads within AI shopping
Amazon’s restrictions on outside tools do not mean it is rejecting automated shopping. It offers Alexa for Shopping, which combines product advice, comparisons and purchasing functions. Amazon said in a June advertising announcement that the service brought together its Rufus shopping assistant and Alexa+, its personalized AI assistant. It said the combination took place the previous month, and Rufus had been used by more than 300 million customers in 2025.
That company-reported reach measures use of Amazon’s own assistant, not customers delegating purchases to an outside agent. More directly relevant to the advertising question, Amazon says advertisers can participate through sponsored ads and Sponsored Products and Sponsored Brands prompts within Alexa for Shopping.
Existing sponsored-ad campaigns are automatically eligible to serve in Alexa for Shopping without additional setup, Amazon says. Prompts in shopping results and product pages can open a product conversation with the assistant. The company also pitches its ability to track ad impressions through to purchases. This gives advertisers a route into AI shopping within Amazon’s system, although the announcement does not compare those ads with orders originating through outside agents.
AI therefore creates competing possibilities: it can shift selection away from a retailer’s interface, or become another advertising surface the retailer operates. Walmart also retains its own assistant, Sparky. The contest concerns who supplies the assistant and controls its connection to the store, as much as whether shoppers use AI at all.
Other retailers illustrate the intermediate choices. An October industry analysis, drawing on Wall Street Journal reporting, said a QVC Group technology executive described QVC and HSN allowing agent purchases and supplying catalogs to OpenAI and Google. It also described Gap deciding browsing and purchasing access case by case. Tapestry, the company behind Coach and Kate Spade, allows purchases through Google channels while prohibiting agents from completing purchases on its brands’ sites over fraud and inventory-hoarding concerns. Browsing access, recommendations and permission to spend are distinct decisions.
The next test is incremental sales and retained influence
Walmart’s immediate test is delivery of the profiles and gateway it announced, followed by evidence that they reduce the failures it describes in browser-based shopping. Amazon’s commercial test is whether its own assistant retains shoppers and advertisers as outside agents compete to handle discovery and purchasing.
The missing economic evidence is whether agent-originated orders are additional sales, how often they bypass sponsored recommendations, and whether lost advertising income outweighs the value of those purchases. Adoption measures also need to distinguish asking an assistant for advice from authorizing it to spend. Until those results are available, a retailer’s willingness to admit agents—or its decision to restrict them—does not settle the advertising threat.
PwC finds growth in a subset of AI-exposed junior job postings demanding senior skills, alongside pressure on entry-level opportunities. Workplace research suggests AI can support learning, but leaves employers to demonstrate how recruits acquire independent expertise.
CoreWeave bonds fell after the election of candidates favoring data center restrictions. Texas governor-elect Gina Hinojosa wants to extend an existing approvals pause until lawmakers pass protections, adding delivery uncertainty for an industry whose expansion depends on power and financing.
Apple reportedly cut some engineers’ daily AI allowances from $300 to $200 in October and is weighing deeper restrictions. The limits tighten control over model use, but unused allowances make the savings uncertain.
Factory is acquiring coding-agent collaboration platform HumanLayer after raising $200 million in September. It plans to integrate shared tasks and planning tools, while HumanLayer promises open source, refunds and at least 30 days of continued service.
A minister identifies Google-commissioned research behind the action plan’s £400 billion AI claim. The government’s central estimate is £120 billion, but different methods and adoption assumptions prevent a straightforward forecast-cut comparison.
Research has separated language models’ detection of altered internal activity from their ability to identify its content. Mechanistic studies and skeptical controls leave the interpretation disputed—and one intervention that improved detection made other measures of self-knowledge worse.
The IEA maps 134 policies governing data-center integration, arguing that credible project queues, coordinated tariffs and flexible connections can protect consumers while accommodating AI growth. Many measures remain too new to judge.
Three watchdog groups seek to halt Project Meridian, alleging that its contractor-heavy membership and closed meetings violate federal advisory committee law. MITRE’s invitation for outside ideas complicates the dispute over public participation.
Meta is seeking up to €9 billion in its first euro bond sale, drawing record orders while offering wider spreads than Alphabet. Its advertising business and large liquid reserves cushion an AI investment push that has overtaken quarterly operating cash flow.
Pangram is expanding its text-detection business into images and video. Browser checks and developer access broaden its reach, while company benchmark scores leave real-world error rates and costs unresolved.
About two dozen restrictive data center measures appeared to pass in initial returns. Texas governor-elect Gina Hinojosa’s promised construction pause adds a state-level constraint, though the capacity ultimately delayed remains uncertain.
Riot Platforms says its 15-year Corsicana lease with Anthropic will generate about $28 billion in revenue. The tenant owns and operates the compute, while Riot develops power infrastructure and funds module reimbursements through debt or credits against future rent.
Sonos plans limited U.S. early access for its conversational assistant, with a reported November 6 start. Users can return to older voice controls—a choice that also separates local audio processing from the new assistant’s cloud processing.
Tesla, SpaceX and AI infrastructure shares fell after Democratic election wins, while broader stocks rallied. Texas governor-elect Gina Hinojosa’s proposed data center pause puts a major construction pipeline in focus, but Republicans retain control of the Texas Legislature and federal agencies.
Capital One Ventures leads Sentra’s $150 million Series C and gains a board seat as the company expands AI-agent data controls. Sentra reports roughly 100 major enterprise customers and revenue in the tens of millions, but exact revenue, valuation and comparative costs remain undisclosed.
Google agreed to support Anthropic’s lease obligations at TeraWulf’s Kentucky campus, strengthening its bid for project financing. The support starts after delivery milestones, while debt terms and a guaranteed construction price remain unsettled.
OpenAI plans to acquire Pierre Computer and integrate its Code Storage service into Codex. The purchase would extend an existing supplier relationship into code infrastructure, but its terms, customer economics and relationship to OpenAI’s own Pando system remain unresolved.
OpenAI says India leads ChatGPT’s markets by weekly active users. Earlier free offers, strong app engagement and planned enterprise deployments explain the stakes, but the company has not disclosed comparable paid-user or revenue figures.
A national exit poll found broad opposition to new local data centers and a Democratic tilt among voters who prioritized AI. But the issue ranked below four other tested concerns, industry-backed candidates also won, and the next fight is over siting and power costs.
