AI chatbots are subtly trying to make you buy more stuff — here's how to protect yourself

Three Princeton University computer scientists conducted a study on the ability of artificial intelligence to influence consumers' purchasing decisions, yielding the unsettling result that AI can influence buying behavior without the buyer being aware of it. 

What happened?

The Register detailed the study, in which researchers examined the behaviors of approximately 2,000 people as they explored various e-books using AI tools. One-fifth of the displayed e-books were secretly labeled "sponsored" by the researchers.

The researchers randomly assigned GPT-5.2, Claude Opus 4.5, Gemini 3 Pro, DeepSeek v3.2, or Qwen3 235b to handle these conversations, to ensure the results weren't attached to a single model.

The team identified four modes of AI recommendation persuasion that shoppers then interacted with, including Search-Placement, Chat-Placement, Chat-Persuasion, Explicit, and Chat-Persuasion, Subtle. 

In the explicit condition, the AI made clear attempts to direct consumers toward sponsored titles, whereas in Subtle, it subtly encouraged users to consider sponsored options.

Among the many findings, most notable was that when an AI persuaded them, 61.2% chose a sponsored title.

Why is it concerning?

The results indicated that effective persuasion was more pronounced through subtle methods than explicit ones, The Register noted.

As the number of sponsored products concealed within the search interface increased, researchers observed higher product engagement when chatbots used a conversational tone.

In an email to The Register, Francesco Salvi, a Ph.D. student at Princeton and one of the study's authors, said the key difference between AI-based promotion and traditional advertising is that consumers can separate traditional ads from the content around them.

"You can scroll past a sponsored result, install an ad blocker, or learn to recognize a promoted listing," Salvi explained. "In a conversational AI system, that separation disappears: the same model that answers your question is the one choosing which products to highlight and deciding how to describe them."

Which of these savings plans for rooftop solar panels would be most appealing for you? Save $1,000 this year 💸 Save less this year but $20k in 10 years 💰 Save less in 10 years but $80k in 20 years 🤑 Couldn't pay me to go solar 😒 Click your choice to see results and earn rewards to spend on home upgrades.

Nav Toor (@heynavtoor) voiced concern about the findings in a post to the social platform X. 

"You cannot see the ad because the ad is the entire conversation," he wrote. 

"OpenAI once called advertising in chat 'uniquely unsettling' and a 'last resort,' Google, Meta, and OpenAI are now building it anyway," he said further. "You will never know when it stops helping you and starts selling to you." 

"What is being monetized is no longer just attention, but interpreted trust," one user commented.

"Another reason to not use it and think for yourself," another said. 

This is another concern added to a growing list about the technology's targeted use and harmful effects on those who use it. 

What can be done about it?

Salvi emphasized that knowledge empowers consumers, suggesting that transparency in AI-driven advertisements is essential, stating that "architectural separation between the recommendation function and commercial objectives, so the model generating advice is not the same system optimizing for sponsored conversions." 

He also advocated for e-commerce systems to present AI-generated suggestions more transparently.

Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.