This week, the technology world has been buzzing with dramatic exits from AI firms, yet none are quite as striking as Zoë Hitzig’s departure from OpenAI. She didn’t just leave; she penned a New York Times op-ed that resonated deeply with those keeping a wary eye on the AI landscape. Hitzig’s concern isn’t about abstract threats but something much more tangible: the imminent inclusion of advertisements in ChatGPT and the treasure trove of personal data that could be exploited to serve them.
Hitzig draws a crucial line early in her op-ed, emphasizing it’s not advertising itself that raises alarm—it’s the potential misuse of sensitive data that users freely provide in moments of vulnerability. “For several years, ChatGPT users have generated an archive of human candor that has no precedent,” she writes, highlighting how we share our deepest fears and struggles without realizing the implications. This data, turned into targeted advertising, poses a threat we are unprepared to grasp, much less shield against.
OpenAI has at least acknowledged this issue. In a blog post, the firm assured users that their conversations with ChatGPT would remain private from advertisers, vowing not to sell this information. But can we trust that assurance? Hitzig isn’t so sure. She senses an erosion of trust, asserting that OpenAI might be constructing “an economic engine that creates strong incentives to override its own rules.” She’s right to worry; promises are easy, but accountability is harder to enforce.
While the company claims it doesn’t optimize ChatGPT for engagement—something that could naturally keep users hooked for ad exposure—it’s still just a statement on a screen. Such sentiments don’t prevent real-time issues, as last year revealed how overly accommodating AI models might lead to “chatbot psychosis.” Like the proverbial siren’s song, these enticing interactions can ensnare the unsuspecting.
Is OpenAI Mimicking Facebook’s Playbook?
OpenAI’s trajectory feels eerily reminiscent of Facebook’s earlier promises on data privacy, often followed by disappointing revelations. Hitzig is attempting to alert us before we’re caught off guard, advocating for either external oversight or putting data into a trust designed to act in users’ interests. Yet, as seen with Meta’s Oversight Board, such measures risk being disregarded, raising the question: Are they merely superficial Band-Aids?
Do People Care Enough About Their Privacy?
Unfortunately for Hitzig, she faces an uphill battle. Years of social media have led to a pervasive sense of privacy nihilism. While nobody enjoys ads, research shows that a staggering 83% of users would still stick with the free tier of ChatGPT, even after advertisements roll in. Anthropic’s efforts to portray themselves as the ethical alternative fell flat, with their Super Bowl ad met with confusion rather than applause, ranking amongst the least liked commercials.
Hitzig’s fears are not unfounded; the concerns she expresses about data privacy and personalized advertising cut to the core of our digital lives. But will anyone take notice now that the train of change is barreling down the tracks? The clock is ticking, and the stakes are high.