Millions of people opened their email and found an artificial intelligence they never requested summarizing their private correspondence. This was not an isolated event. It was part of a coordinated architectural shift executed by the largest technology platforms on the planet.
Google’s Gemini assistant appeared in Gmail unprompted, mirroring the company’s earlier move to place non-optional AI Overviews at the top of its search results. Meta followed the same playbook, embedding its Meta AI chatbot as an irremovable fixture within Instagram, WhatsApp, and Messenger. The internet is beginning to look different for every single user, a bespoke reality of tailored ads and unique product prices shaped by conversations with chatbots. There is no off switch. The tech industry is building a personalized internet for you, but it has declined to ask for your permission.
Analysts at AI ethics firms like Hugging Face note the inversion of control. “These tools are sold to us as more powerful, but we have less say in things,” researcher Sasha Luccioni stated. The burden shifts to the user to navigate complex and obscure menus to opt out of systems they never opted into. This strategy directly conflicts with public sentiment; a Pew Research Center survey found Americans are more concerned than excited about AI, with a majority demanding more control. Tech companies insist they are focused on creating the best “agent” to empower people. This is PR speak. The true objective is economic.
A New Engine for Advertising
The immense operational cost of generative AI models cannot be sustained by consumer subscription fees alone. The real business model is a fundamental rewiring of the digital advertising economy. For years, regulators have dismantled the infrastructure of third-party tracking that allowed companies to follow users across apps and websites. In the conversational interface of a chatbot, companies found a new, richer data source: direct, voluntary disclosure.
When a user interacts with an AI assistant, they reveal their intentions, interests, and vulnerabilities far more explicitly than a simple keyword search ever could. They ask for gift ideas, plan vacations, and discuss health conditions. Web search queries are already rising industrywide as chatbot-powered search engines encourage more follow-up questions. More questions mean more data. This is not about building a helpful assistant. (It is about building a more efficient harvesting mechanism).
OpenAI announced it would begin showing ads in ChatGPT’s free version, based on user conversations. A Google executive dismissed the move, stating Gemini would remain ad-free, yet neglected to mention that Google’s core search product already leverages AI interactions to target the ads that finance the entire company. The conversational interface is the new data mine.
Surveillance Capitalism Recoded
The information collected is more intimate, so the advertising it powers feels more invasive. Both Meta and Google are deploying systems that use AI to automatically construct and tailor ad creative. Based on chatbot conversations, an ad system can infer a user enjoys running in the winter and dynamically generate marketing copy emphasizing a specific shoe’s durability in cold weather. The goal, according to consultants at firms like Madison and Wall, is to personalize an ad just enough so the AI’s involvement remains invisible. “I think of all this A.I. optimization as bad plastic surgery,” said managing director Luke Stillman. “You notice the ones that stand out, but you don’t notice the ones that are still standing behind the curtain.”
The logic extends beyond advertising into dynamic pricing. Google recently unveiled an AI-powered shopping tool with retail giants like Target and Walmart. Economic watchdogs at the Groundwork Collaborative label this framework as “surveillance capitalism” designed to compel people to spend more. Information a consumer shares with a chatbot—a personal budget, an impending birthday—becomes a variable in a pricing algorithm. A clothing brand could calculate that a frugal shopper needs a $32 discount to purchase a $120 jacket, while offering the same item at full price to a wealthier user. (Google states it prohibits retailers from inflating prices, but the architecture for differential pricing is being built).
The Illusion of an Alternative
Smaller rivals have positioned themselves as a refuge. Mozilla, maker of the Firefox browser, is building a broad set of controls to manage individual AI features, arguing that a web dominated by a few powerful AI models is a less open web. The privacy-focused search engine DuckDuckGo launched a version of its tool that filters out AI-generated images, after a poll showed nearly 90 percent of its users wanted a non-AI option.
This resistance, however, runs up against a wall of market dominance. It is difficult to avoid the AI baked into Google and Meta products when their services are deeply integrated into daily life. Google commands roughly 90 percent of the global search market. Mark Zuckerberg reported that 3.58 billion people, about 44 percent of the world’s population, use at least one Meta product daily. The user is locked in. “Big companies that have users locked in can get it wrong for a very long time before they ever get fired by a user,” observed Kamyl Bazbaz, an executive at DuckDuckGo. Google’s AI Overviews were widely seen as a poor user experience for two years. People did not leave. They had nowhere to go.