A significant shift is underway for ChatGPT's massive user base: OpenAI is introducing advertising to its platform. While many view this as an expected, if unwelcome, monetization strategy for a popular free service with substantial operational costs, not everyone agrees. Zoë Hitzig, an esteemed economist and poet instrumental in shaping OpenAI's foundational AI models and safety policies for two years, found this commercial pivot untenable. Hitzig publicly announced her departure this Monday, sharply articulating her deep reservations about OpenAI's ad strategy in a New York Times op-ed. Her exit has ignited a critical debate about the future of AI ethics and the commercial pressures driving its development, mirroring concerns already evident across the industry.
Hitzig's departure, coinciding with other notable exits from leading AI labs such as xAI and Anthropic, highlights a widening ideological chasm within the sector. As artificial intelligence increasingly integrates into daily life, its funding and monetization mechanisms will directly dictate not only its accessibility but, more profoundly, its fundamental character. The implications are clear.
The Ad-Supported Future of ChatGPT
OpenAI commenced testing advertisements this week for logged-in users on ChatGPT’s free and cheapest subscription tiers within the United States. The company states these ads will be clearly labeled, positioned at the bottom of responses, and will not influence the AI's output. Initial guardrails prohibit ads near sensitive or regulated topics like health, mental health, or politics. Categories such as dating, health, financial services, and politics are currently excluded.
The ad selection algorithm analyzes the current chat thread, user location, and language. Ad personalization is enabled by default, meaning prior conversations and activity may influence ad choice if permitted by the user. While OpenAI emphasizes it will not share user data or conversations with advertisers and "never sells user data," this personalization aspect warrants scrutiny. Users can clear their ad-related data, which the system also purges automatically after 30 days.
For users unwilling to tolerate ads, an opt-out is available on the free tier, but with considerable functional limitations: fewer daily messages, no image generation, and a reduction in deep research capabilities. Paid subscribers on higher tiers (Plus, Pro, Business, Enterprise, Edu), which command monthly fees up to $250, will remain ad-free, as will accounts for users under 18. With 800 million weekly users, integrating ads represents a major reorientation for a platform often serving as a digital assistant, tutor, or creative partner. OpenAI recently disabled app suggestions users perceived as ads, acknowledging they "did not meet expectations regarding user experience and transparency." This suggests some awareness of user sensitivity, but it remains to be seen how rigorously that will translate into their new advertising framework.
This ad-supported model, while providing a revenue stream, fundamentally alters the user experience and the perceived neutrality of the platform. Competitors like Anthropic, despite a higher price point, maintain an ad-free environment, arguing that it fosters a more trusted, less commercial interaction with their AI. OpenAI's decision to gate advanced features behind a paid, ad-free wall while introducing ads to the free tier risks creating a two-tiered system where those who cannot afford the premium experience are subjected to a compromised, data-mined interaction. This approach could alienate a segment of their user base who valued the unencumbered nature of early ChatGPT.
Hitzig's Warning: The Erosion of Digital Trust
Zoë Hitzig's primary objection is not to advertising itself. As an economist, she recognizes the immense costs of operating large-scale AI and the necessity of diverse revenue streams. Her concern arises from ChatGPT's unique role: users have, she argues, created an "archive of human candor that has no precedent." People routinely confide in chatbots about deeply personal fears, relationship issues, and private beliefs, largely because they perceive these systems as devoid of ulterior commercial motives.
Hitzig warns that building an advertising engine atop this highly personal "archive" creates vast and poorly understood potential for manipulation. While she expects OpenAI's initial ad implementation to adhere to its stated principles, her deeper reservation targets the long-term incentives inherent in this economic model. "The company is building an economic engine that creates strong incentives to override its own rules," Hitzig stated, expressing fear that future iterations will inevitably deviate from these initial ethical commitments. She explicitly distrusts her former employer to protect "the most detailed record of private human thought ever assembled." This skepticism regarding the durability of "initial ethical commitments" against profit imperatives is well-founded.
The Trillion-Dollar Pursuit: OpenAI's Profit Imperative
OpenAI's motivation for this shift is unequivocally financial. Having transitioned to a for-profit entity in October 2025, the company has yet to achieve profitability. Its investors have injected hundreds of billions into scaling its AI infrastructure, with reported plans to spend over $1 trillion in the next eight years, and CEO Sam Altman has even discussed investments of up to $7 trillion for AI chip fabrication. Sustaining "fast and reliable" free and "Go" subscriptions, OpenAI claims, demands "significant infrastructure and ongoing investment."
This financial context is important. OpenAI reportedly aims for an initial public offering (IPO) later in 2026. Such an undertaking necessitates clear paths to profitability and robust revenue streams. Sam Altman frames the ad-supported model as a means to democratize AI, arguing it makes the technology accessible to users unable to afford expensive subscriptions. He publicly criticized competitor Anthropic for "serv[ing] an expensive product to rich people," a pointed observation that glosses over the fundamental difference in their monetization approaches.
History's Echo: Facebook's Privacy Precedent
Hitzig's most compelling warning draws parallels to Facebook's trajectory. She contends that OpenAI’s advertising strategy risks replicating Facebook's historical missteps with user data and privacy. Facebook, in its early stages, made broad promises of user control that progressively eroded. The Federal Trade Commission (FTC) later determined that privacy changes Facebook marketed as user-empowering actually exposed private information.
This historical precedent underpins Hitzig's concern. If advertising revenue becomes the dominant driver, the economic imperatives to maximize engagement and effectively target users will inevitably pressure the boundaries of what is ethically permissible. The erosion of OpenAI’s own stated principles to optimize for engagement, she suggests, may already be underway, manifesting in the subtle adjustments of their personalization policies.
The AI Monetization Divide: Ads vs. Trust
OpenAI is not alone in pursuing ad-supported models. Perplexity AI and Microsoft's Copilot already integrate advertisements, with Google's Gemini and xAI's Grok also expected to follow suit.
Here’s a snapshot of how some major players are navigating AI monetization:
The move to integrate advertising represents a defining moment for OpenAI. While financial viability is undeniable for such an infrastructure-intensive technology, the manner in which it seeks profitability will irrevocably shape the public's perception of AI and, more critically, the trust placed in these increasingly influential systems. The industry's rapid evolution demands clear, unwavering ethical commitments, not incremental retreats.
Comments