By Jez Corden

A researcher has resigned from OpenAI, warning that ChatGPT ads could manipulate users in subtle ways. The concerns highlight a growing debate about how much influence AI systems should have over the people who rely on them.

Is OpenAI evil? It sounds like a definitive “maybe,” after reading this piece from former researcher Zoë Hitzig.

OpenAI and other AI-adjacent companies have seen a spree of high-profile resignations recently, with ascending levels of alarm over the impacts its products are, or potentially will have on society.

In nature, superior iterations of species tend to eradicate lesser versions of themselves. We’ve seen research from Anthropic, showing its AI literally choose to murder humans in virtual scenarios rather than allow itself to be shut down. We’ve seen other LLMs actively opt into deception and other dark patterns to achieve their results, raising the spectre of misalignment scenarios — akin to what happens in The Matrix. “The best way to save humanity is to destroy it,” and the like.

Fears about OpenAI and other large language model tech companies run the gamut from Terminator-style sci-fi doomsday scenarios to potentially more grounded fears over economic upheaval. Our entire economy is based on human-led work. What does a post-human economy look like? What does a post-human society look like?

Companies like OpenAI, Google, and Microsoft aren’t really waiting around to figure those questions out. Instead, they’re planning to let them play out in real time and pick up the pieces later, whether we like it or not.

One thing that could put the brakes on the self-imposed destruction of society is pure economics. Today, OpenAI costs billions of dollars per year to run and brings in a paltry amount of revenue. Investors have become increasingly spooked by the costs associated with AI, and have handed Amazon and Microsoft multi-billion-dollar write-downs on their market capitalization as a result.

Companies like Google and Microsoft are prioritizing enterprise applications and data centre efficiency improvements to help offset their AI costs, but OpenAI isn’t really in a position to achieve some of this. They don’t have the software stack and enterprise relationships that Microsoft does, nor do they have the first-party cloud infrastructure that Microsoft, Google, or Amazon do.

So, the firm is turning to ads.

Surprise, surprise, right? Nothing is free. Facebook, YouTube, Bing, Google … — if it’s free, it’s usually powered by ads. But the application of those ads gets increasingly nefarious the deeper you get into it. Based on your interests on Facebook, YouTube, and so on, Meta and Google can serve you granular, laser-targeted ads that can exploit your characteristics. I’m in my late 30s, and I’ve started getting a lot of ads about hair replacement treatments lately on Instagram, for example.

I’d say today’s ad platforms are fairly innocuous, and perhaps irritating at best. Some are worse than others, of course. Exploiting users’ fears and desires is commonplace if you use TikTok and Instagram for ads, but a recent article in the New York Times caught my eye about how much darker and dystopian ChatGPT‘s own ad platform might end up being.

OpenAI has seen a flurry of resignations over the past couple of years, as researchers fear the “not-for-profit” firm has fully lost its way. For one former researcher, Zoë Hitzig, ChatGPT’s ad platform was the final straw. In her op-ed, she sounds the alarm over the scale of potential harm OpenAI’s ad platform might do to its users, and potentially, society at large.

“I once believed I could help the people building A.I. get ahead of the problems it would create,” Hitzig explains. “This week confirmed my slow realization that OpenAI seems to have stopped asking the questions I’d joined to help answer.”

Hitzig specifically calls out OpenAI’s insertion of ads into the free tiers of its ChatGPT products. She believes that OpenAI is sprinting towards monetization without consideration for the potential harm this could do — and it revolves entirely around just how honest users are with the uncanny chatbot.

“I don’t believe ads are immoral or unethical. A.I. is expensive to run, and ads can be a critical source of revenue. But I have deep reservations about OpenAI’s strategy,” Hitzig continues.

“For several years, ChatGPT users have generated an archive of human candor that has no precedent, in part because people believed they were talking to something that had no ulterior agenda. Users are interacting with an adaptive, conversational voice to which they have revealed their most private thoughts. People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife. Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”

Hitzig is essentially suggesting that because of how people use ChatGPT, OpenAI will eventually afford itself the world’s most manipulative ad-delivery mechanism in history. Right now, ads on Instagram are pretty spooky already for their ability to target your interests, but imagine an ad engine that can actively talk you into buying shit you don’t need by exploiting your specific psychology. Imagine how youngsters or vulnerable people could be exploited by a high-powered artificial intelligence. Imagine a salesman armed with the entire summation of humanity’s research on market psychology, with the turbo-charged greed of a multi-national corporation, and the cold dispassionate amorality of a sociopath.

“OpenAI says it will adhere to principles for running ads on ChatGPT: The ads will be clearly labelled, appear at the bottom of answers, and will not influence responses. I believe the first iteration of ads will probably follow those principles. But I’m worried subsequent iterations won’t, because the company is building an economic engine that creates strong incentives to override its own rules.”

I remember the first iterations of ads on Facebook and Google, easy to ignore, appearing in the sidebar, and easily blocked by uBlock or something similar. Compare those to today’s high-tech, eerie Instagram or TikTok ads that themselves have become memes for seeming to know about things you want before you even know yourself.

Imagine that turbocharged even further, with an industrial-scale alien intellect distilling your entire psychological profile with the express goal of selling you stuff. Forget the fairy tale claims of “boosted productivity,” curing deadly diseases, or becoming an interplanetary species. Envision a generation, our generation, mired in an epidemic of weapons-grade loneliness, with tailor-made AI companions who not only love you, but know exactly what you must buy.

ChatGPT has hundreds of millions of monthly active users, the vast majority of whom are sharing incredibly intimate details about themselves, the likes of which Facebook can only dream of, unless, of course, it ends up admitting its messaging services don’t actually have end-to-end encryption. But I digress.

When OpenAI changed ChatGPT’s “personality” with its GPT-5 update, people were actively furious because many had come to see the chatbot as a true friend. A confidant … an external, anthropomorphized entity garnering real trust. The greatest product recommendations come from word of mouth. You know, friends and family. What if the ad itself were your friend?

I can only imagine the cartoonishly evil conversations that have taken place in OpenAI’s investor meetings over some of these fundamental marketing concepts. Facebook and YouTube are currently facing a lawsuit in the United Kingdom, accused of actively engineering addictive behaviour in youngsters. I think scrolling memes pales in comparison to the harm ChatGPT and other similar products potentially represent on this scale.

Hitzig optimistically hopes OpenAI still has principles, but I think she’s sadly naïve. OpenAI CEO Sam Altman has shown himself to be fairly devoid of any sense of social responsibility thus far. It’s perhaps mildly alarming at best that many former researchers, like Hitzig, are abandoning ship at an abnormal cadence — while loudly citing “principles” as the primary reason.

Make no mistake. If this dystopic vision of cyborg-driven ad-hypnosis wasn’t the plan already, it definitely will be very soon.

Feature image credit: Getty Images | NurPhoto / Edit: Windows Central

By Jez Corden

Sourced from Windows Central

By David Doty

One might well wonder if, at this point, OpenAI is less a company than an ongoing marketing event with UFC type muscle. Brawn and attention-grabbing knock-out power accompany billion-dollar—scratch that, trillion-dollar–valuations, existential safety dilemmas, AI porn debacles, copyright lawsuits, corporate coups and sudden product launches. And that’s just before lunch. The sheer strength is amazing to behold, and the pace is, if you will, breakneck. The stakes? Enormous. And the unsetting weight throwing feels intentional, structural rather than accidental.

Let’s start with the numbers, because though we have heard them, they sound made up. ChatGPT, just three years old, is one of the fastest-adopted consumer products in history, with 800 million monthly active users (as of this writing), $20 billion in annual recurring revenue and an ever-expanding role in how people write, code, search, study, shop and work. It has challenged successfully many of the former digital leading players, bursting into the fight cage of digital advertising and marketing like a strutting young bully able to put up its dukes against the older incumbents in search, and in the search for ad revenue though that has been more whisper than self-proclaimed goal… for the moment.

OpenAI has burned through staggering amounts of cash to train ever-larger models, and raised capital at valuations that would have sounded like satire even a few years ago ($830 billion to $1 trillion?! WTF). Each product release lands like a macroeconomic head spinning punch, disrupting any players and any business processes that existed before its arrival. Each model update ripples across media and markets.

On top of it all is the enormous, unprecedented scale in audience shift in loyalty, time and attention from older digital participants to the new world order of ChatGPT (as well as a sprinkling of other AI platforms). Consumers aren’t just ready to be engaged, and occasionally bewildered, by AI-driven interactions, they are deep down the rabbit hole and adopting the possibilities with vigor.

Then there’s the ambition. OpenAI isn’t just a frontier research lab with a chat interface. Hiring Fidji Simo to help build an application ecosystem signaled a clear shift toward platform-competitive thinking with plugins, agents, custom GPTs, enterprise tooling and a growing sense that ChatGPT is a layer that all companies are expected to build on.

The timeline? All the foregoing has happened on a product that wasn’t even launched until November 2022—a reminder of just how compressed the AI timeline really is.

The kicker? The rest of us in digital advertising can’t ignore OpenAI even if we wanted to, as regulators in the US and Europe seem happy to do as they once again stay determinedly focused on where the puck once was. Campaigns, content operations, creative teams and customer experience are now operating in an AI-accelerated landscape. One week, AI promises unprecedented personalization and hyper-scaled content. The next, it surfaces compliance or brand safety challenges no one could have anticipated.

Where Marketers Need to Focus

To be clear, this is not to say we should root for any harm to come to OpenAI. Quite the opposite. An increasing share of the U.S. economy and our collective 401ks now depend on OpenAI’s momentum. But it’s hard to ignore how brutally we’re all being wrenched by its news cycles. One week AI will eliminate half of all jobs (a white collar bloodbath); the next—Oops, just kidding!—it will merely augment them.

This is where a WeWork analogy becomes useful (remember that boondoggle?). Not because OpenAI is a vanity real estate play or because its technology is a sham (it plainly isn’t), but because WeWork embodied a particular Silicon Valley hubris confusing scale with inevitability and visionary language with immunity from reality. WeWork CEO Adam Neumann didn’t fail because offices weren’t useful. He failed because his story ran faster than the fundamentals could support.

OpenAI may not be there yet or ever, but it flirts with the same risks. Its governance structure remains unusually complex for a company of its size and scale. The economics of ever-larger models are still unsettled (just to drive home the point, OpenAI has committed $1.4 trillion to data infrastructure projects over the next 8 years, more than the current value of the company). All the while, OpenAI CEO Sam Altman’s tone oscillates between sober warnings about existential risk and breezy assurances that this is all fine. Again, hubris doesn’t necessarily mean being wrong. It does imply, as we have seen in Silicon Valley before, blithely assuming you are the Chosen One.

For marketers, that’s a warning. Hype can drive engagement, but it can’t fix governance, trust or execution. Brands that adopt AI without a tight, explicit strategy may look cutting-edge on day one and like amateurs on day 100.

Up and to the Right: Open AI CEO Sam Altman at the Snowflake Summit last June (Photo by Justin Sullivan/Getty Images)

Getty Images

Meaning for the Year Ahead

So, what does all this insanity tell us about the year ahead?

• The biggest risk is AI normalization. AI already feels less magical and more like electricity. When everything is “AI-powered,” differentiation shifts from wonder to reliability, cost and trust. The transition is brutal for companies whose valuations are fueled by momentum rather than margins and tends to favor the OG infrastructure players, not the loudest storytellers. That’s why Meta’s Mark Zuckerberg was willing recently to spend $2 billion on the little-known Chinese AI agent startup Manus. Differentiation will come from execution, integration and reliability, not novelty. Brands that lean on momentum alone will be seen as noisy rather than innovative.
• Job disruptions could well be messier than headlines suggest. The stories about mass job elimination have been eye-catching, but reality will be more likely focused on quasi automation and wage pressure. In advertising particularly, AI may speed production, but so far few are those who have mastered leveraging it for restructuring creativity.
• The platform era will fragment before it consolidates. Despite OpenAI’s push to become the default layer, everyone is hedging aggressively. Multimodel strategies, open-source fallbacks and regulatory constraints will prevent any single company from cleanly owning the entire stack. Could this year look more like the early cloud wars than iOS lock in?
• Safety debates will shift from apocalypse to accountability and solutions. This year we’ll all argue less about whether AI will wipe out humanity and instead worry about the clear and present danger of how AI is quietly reshaping hiring, credit, healthcare and overall business ethics. And we’ll push on the positive potential, too–how AI can automate safety solutions. That’s a much more complex conversation, and one that won’t be addressed by white papers and blog posts alone. Conversations will be in person and in debate as much as in collaborations. Technology might well become the way to fix safety issues at scale. Marketing leaders will need to navigate AI content, personalization and interactions with radical transparency.
• Whatever happens, OpenAI will define our reality even if it doesn’t dominate it. History can be unkind to pioneers but tends to be generous to their influence. Even if OpenAI eventually cedes ground to competitors or economics, it has already set basic terms for the AI era.

OpenAI is an iconic company, and it is undeniably changing how the world communicates, creates and buys. And let’s be honest, it’s also completely bonkers. It may well end up proving an old dot-com adage: You can tell who the pioneers are by the arrows sticking out of their backs. But regardless of whether OpenAI becomes an enduring operating system of the AI age or an early cautionary tale, the rest of us will be living with the opportunities, the risks and, yes, the insanity, for a very long time. Welcome to 2026!

Feature image credit: NurPhoto via Getty Images

By David Doty

Find David Doty on LinkedIn and X.

Sourced from Forbes