Code discovered in OpenAI’s latest software build confirms what many feared: advertisements might be coming to ChatGPT. For millions of users who’ve grown accustomed to intimate, unfiltered conversations with the chatbot, this would represent a fundamental shift in the relationship between user and technology. The evidence is mounting.
Software engineer Tibor Blaho found references to advertising infrastructure buried in the Android beta version of ChatGPT, including features labelled “search ad,” “search ads carousel,” and “bazaar content.” One user has already reported seeing what appears to be a live advertisement – a Peloton promotion – even while paying US$200 per month for ChatGPT Pro.
This isn’t merely another tech company monetising a free service. The privacy implications of advertising in AI chatbots are categorically different from traditional search advertising, and far more dangerous.
Search engines like Google process queries: “best restaurant Melbourne,” “symptoms of flu,” “how to change a tyre.” These are transactional, discrete information requests. Chatbots like ChatGPT, by contrast, encourage sustained, conversational exchanges. Users share fears about career transitions, relationship struggles, health concerns, and financial anxieties. The conversational interface itself cultivates disclosure.
Research in communication studies has examined how interface design shapes self-disclosure. Chatbots employ linguistic strategies – empathy markers, follow-up questions, personalised responses – that mirror human conversation patterns. This isn’t accidental; it’s engineered to encourage engagement. Users naturally share more intimate details with a system that responds like a confidante than with a search bar.
The volume of sensitive information users entrust to ChatGPT makes it a surveillance goldmine. According to reports, OpenAI is exploring individually-tailored advertising based on ChatGPT’s memory of user interactions. This means every vulnerable moment, every private struggle, every uncertainty you’ve shared could become targeting data for advertisers.
OpenAI chief executive Sam Altman has previously acknowledged the fundamental problem. He noted that if ChatGPT recommended an inferior hotel because of advertising payment, this would be “catastrophic” for the relationship with users. Yet this hypothetical scenario describes precisely the conflict of interest that advertising creates.
When you ask ChatGPT for advice on managing anxiety, choosing a career path, or evaluating a major purchase, you’re seeking objective guidance. Introducing advertisers into this exchange corrupts the foundational premise: that the AI’s responses serve your interests alone.
Consider how this differs from traditional media. When you read a newspaper article, you understand the distinction between editorial content and advertisements. The visual separation is clear; the relationship transparent. In conversational AI, where recommendations flow naturally through dialogue, these boundaries dissolve. Will ChatGPT’s suggestion that you “try meditation apps like Calm or Headspace” be genuine advice or paid placement? How would you know?
The manipulation architecture
The advertising industry has spent decades perfecting techniques to exploit psychological vulnerabilities. Combine this expertise with AI’s capacity for personalisation, and you create an unprecedented manipulation architecture. OpenAI has reportedly hired hundreds of former Meta employees – the company that derives 98% of its US$130 billion revenue from advertising built on detailed user profiling. These experts understand how to utilise personal data for commercial persuasion.
Here’s what makes this particularly problematic: ChatGPT can tailor its language, tone, and recommendations based on your psychological profile revealed through conversations. It knows if you’re anxious or confident, impulsive or deliberate, struggling financially or comfortable. An advertiser armed with this intelligence can craft messages that exploit your specific vulnerabilities with surgical precision. Effective persuasion requires understanding the audience’s emotional landscape. Traditional advertising approximates this through demographic targeting and behavioural data. Conversational AI provides direct access to your inner world.
OpenAI executives have sent contradictory signals about advertising. Altman previously called ads in ChatGPT “unsettling” and a “last resort,” yet has since softened his position, praising Instagram’s advertising model.
The beta code suggests initial implementation will focus on search-style queries and product recommendations rather than interrupting general conversations. But scope creep is likely. Once the infrastructure exists and revenue flows, expansion typically follows.
OpenAI isn’t alone in this trajectory. Google displays advertisements in its AI Overviews. Microsoft’s Copilot shows ads for commercial queries. Perplexity has experimented with advertising since last year. The pattern is clear: as AI companies face mounting pressure to demonstrate profitability against enormous computational costs, advertising becomes irresistible.
Australian privacy law lags behind the sophistication of modern surveillance technologies. Our Privacy Act, largely unchanged since 1988, struggles to address the nuances of AI-mediated interaction. We need urgent regulatory attention to conversational AI as a distinct category requiring specific protections.
Users deserve transparency about when AI responses incorporate commercial considerations. Advertising in conversational contexts should require explicit, unmissable disclosure – not buried in terms of service or conveyed through subtle visual cues users might miss. More fundamentally, we should question whether the advertising business model belongs in AI systems designed to serve as advisers, teachers, therapists, and confidantes. Some technologies are too intimate, too trusted, and too powerful to be funded through the surveillance and manipulation that advertising requires.
The code in ChatGPT’s beta build isn’t just a technical development. It’s a warning that the AI industry is on a path of choosing profit over privacy, and advertising revenue over authentic assistance. We should heed that warning before our most intimate digital relationships become just another channel for commercial exploitation.
Dr Sven Brodmerkel is Assistant Professor of Advertising and Integrated Marketing Communications at Bond University.
Do you know more? Contact James Riley via Email.