While investments in the AI industry are breaking new records, large cracks in the public perception of the AI industry have started to appear over the last year. Concerns about AI have moved from a vague “Will AI destroy us?” to more down-to-earth issues: From general annoyance, over environmental concerns to the realisation that the industry might not have our best interests in mind. Reality is creeping into the success story.
In the first part of this article, we will examine various aspects of this backlash. Then we will consider what it means for the industry, for the industry’s relations to governments and regulators — and for everyone else.
The current AI summer started almost like the previous two: With research: The companies we now know to be the juggernauts of the AI industry, OpenAI and Anthropic, were both founded as “safety and research labs”, though little of that remained.
But there are important differences from the previous AI hype cycles. While the AI summers of the past fizzled out because expectations were not met and little commercial practical use was found, the current blitz scaling economy does not concern itself with such mundane problems: AI is here to change the world, as the message goes, “and you will like it”.
But then, a few weeks ago, the CEO of Nvidia, Jensen Huang, took the stage and telling everyone “not to be so negative about AI” (Landumore 2026). Only a few days before, Satja Nadella asked people not to call Windows 11’s sparkling new features “AI slop” (Bort 2026). A remarkable change from only a year ago, when optimism promising a bright future for humanity was the prevailing narrative.
Since 2022, when ChatGPT publicly launched and awed most of the public, we went from “AI will cure cancer and create a paradise” to the rather defensive position where tech CEOs have to beg to give their products a chance.
But something else happened in recent weeks: OpenAI announced it will be closing down Sora, its AI video generator and attempted at a social network, and cancelled the “adult mode” for ChatGPT. Meanwhile, Oracle and OpenAI quietly shelved plans for a new data centre in Texas, reportedly over financing issues (Rosner-Uddin, Hammond 2026).
Despite all the excitements about the allegedly “rapid progress of LLMs”, “agentic AI” and its lofty promises, the commercial uptake of AI services has been lagging. Meanwhile, the AI industry paints a picture of “infinite demand”. After the failure of rapid adoption in the lucrative enterprise market, the AI industry has since eyed the consumer and consequently overdid it: Forced product integration such as in Zoom, Slack or various messengers, often vehemently nagging people to pay for features they did not ask for. The problem is not so much the features themselves, but the brute force by which it is presented to the users.
Similarly, the flood of AI slop that has been filling platforms like Instagram, YouTube and other social media sites has caused some users, who originally came to those platforms to connect to others, to question those platforms themselves. The omnipresence of AI generated disinformation on those platforms has aggravated the situation.
This has reached the point where dissociating a business with AI has become a marketing strategy: Recently, music publishing platform Bandcamp announced, to much approval, that it will be “fortifying [its] mission by articulating our policy on generative AI, so that musicians can keep making music, and so that fans have confidence that the music they find on Bandcamp was created by humans.” (Bandcamp 2026)
Meanwhile, game developer Umamusume was forced to apologise after announcing the use of AI art in its games, says it will not use generative AI “without prior notice” (Eurogamer 2026). And when British Museum posted AI-generated social media content, the backlash caused its social media team to delete the post within hours.
When enterprise adoption is lacking and consumers are not inclined to pay up, there is always advertisement. Sam Altman, CEO of OpenAI, once described adding advertisement to ChatGPT as “the last resort”, a point OpenAI has arrived at much earlier than many, likely including Altman, expected.
But this also sends another message: Advertisement on the internet has always been a way to monetize a product that otherwise is not profitable, such as web search, cheap news content or social media. Adding ads to chatbot signals that generative AI might not be a product people want to pay for itself, and by doing so OpenAI throws its own product into doubt.
In the reverse, AI generated ads are now widespread, and while being a curiosity at first, have also widely met with rejection: “[…] There is nothing innovative about it. And given how much slop there already is in the world, the whole thing feels more like a threat than a promise of good things to come. […] This latest crop of gen AI ads was, in part, aiming to normalize the technology via onslaught.“ (Pulliam-Moore, Weatherbed 2026)
Meanwhile, in the gritty every day impact of the AI boom is felt by many has let to a new form of resistance: May it be in Memphis, where Elon Musk powered his giant data centres with unlicensed gas turbines or elsewhere, people have started to take note of the environmental impact of the industry.
Last week it was reported that a farmer in Kentucky rejected tens of millions offered by an undisclosed AI company to sell her land for a new data centre (Rowe 2026), explicitly pointing out the dangers to the environment and the community. Meanwhile, “[…] at least fourteen states are home to towns or counties that have implemented moratoriums: a complete pause on data center development.” (Raddock 2025).
While often labelled as NIMBY politics by the industry, the resistance against AI data centres appears more fuelled by consciousness of climate change, energy cost and impact on the quality of life that has grown over the last decade. And people have started to ask themselves if it is worth living with a water shortage for Elon Musk to generate millions of nonconsensual sexual imagery (Chun 2026).
As if this weren’t enough, there’s the personal dislike many feel towards the industry’s most prominent figures. Peter Thiel’s transhuman aspirations. Palantir’s opaque entanglement with law enforcement, led by a CEO who casually discusses how his products will “occasionally kill people” (Knowles 2026), or Elon Musk flirting with the far-right and white supremacy.
There is OpenAI’s suggestion of public bailouts, Amazon Ring is painting a dystopian future of surveillance in a Superbowl ad, all while Jeff Bezos plans to “buy up production facilities to automize them” (Predin 2026). Or Sam Altman comparing the energy requirements of AI to that of humans. (Berger 2026)
The distain for culture, the harvesting of artistic work people are emotional invested bundled with the smugness of CEOs pushing AI into areas they have no domain knowledge about is aggrevating the perception of arrogance. And the irony of promising jobs and economic growth while publicly claiming its own technology will replace human jobs in all aspects of life is not lost on the consumer.
A recent poll in the US found that people now hate AI even more than they hate ICE (Yildirim 2026). In the current political climate, that is a strong sentiment.
Rarely has a new industry burned through so much good will in such a short time. This is both a chance to reign in the industry and demand more transparency, safety and accountability, but also a danger of facing a cornered industry that will fight to survive.
“Lastly, lawmakers everywhere saw the backlash, and many joined in. These are the people thinking about and discussing the AI laws that will be enacted in the short, medium, or long term. […] The public outcry and negative reactions have influenced them, and they might, sooner or later, materialize in concrete legal provisions.” (Jarowsky 2026)
Now, one might expect that the AI industry, like so many companies in the Dotcom bubble, could just slowly run out of money, consolidating the market and scaling back to more realistic expectations, while some hyperscalers go the way of Nortel.
The massive investments (partly based on debt) require a wide adoption and the willingness of the consumer to buy into the products. But the more aggressively AI features are marketed, the reluctance will only grow.
For the AI industry, too much is at stake: Hundreds of billions of debt are linked to AI. The “bet on AI” When consumers reject the product, companies turn to governments: By embedding themselves deep into the military, health services and government surveillance.
It is hence no surprise that the AI industry has dedicated enormous resources to lobbying against AI regulation and public oversight: From SuperPACs in the US to supporting democratic backsliding in Europe, rarely before, and maybe not since the days of the United Fruit Company has an industry attempted to intervene so heavily in politics and society.
As a result, the US government has made it official policy to support the AI industry in its aim for deregulation, even in foreign countries.
Over the last three years the AI industry has seen a gigantic influx of investment and attached itself to many aspects of life. Even some of the most die-hard AI evangelists now acknowledge that we are seeing a technological hype and economic bubble. We might expect that this will end like the Dotcom and Telecom bubble, but the changes brought by the AI hype may last longer and be more severe.
The AI industry can not scale back without undermining its own story. That could mean the only way forward for the industry will focus on achieving the societal and economic changes it needs to survive, a deeply concerning development governments and regulators should not overlook.
Rarely has there been so much public support for industry regulation. While historically many opposed industry regulation because they feared the loss of jobs, this cycle is somewhat reversed. This is the worst time to give up on regulation, and policymakers should grasp the chance.
References
Julie Bort (2026) Microsoft’s Nadella wants us to stop thinking of AI as ‘slop’, Techcrunch
Frank Landumore (2026) Nvidia CEO Says Everyone Should Stop Being So Negative About AI. Futurism
Kelvin Chun (2026) Musk’s AI chatbot faces global backlash over sexualized images of women and children.
Bernard Marr (2025) The Growing AI Backlash: Is The Revolution Over Before It’s Even Begun?. Forbes
Luisa Jarovsky (2026) AI Backlash as a Regulatory Tool
Victoria Phillips Kennedy. (2026) Umamusume developer apologises after AI studio backlash, says it will not use generative AI “without prior notice”. Eurogamer.
Charles Pulliam-Moore, Jess Weatherbed (2026). AI-generated ads dropped the ball at this year’s Super Bowl. The Verge.
Bandcamp Blog. (2026) Keeping Bandcamp Human.
Josipa Majic Predin (2026). Jeff Bezos Is Targeting $100 Billion To Acquire And Automate The Manufacturing Sector. Forbes
Eric Berger (2026). Sam Altman defends AI’s energy toll by saying it also takes a lot to ‘train a human’. The Guardian
Tom Knowles (2026). ‘Our product is used, on occassion, to kill people’: Inside Palantir, the world’s scariest AI company. BBC Science Focus
Rafe Rosner-Uddin, George Hammond (2026). Oracle and OpenAI scrap deal to expand flagship Texas data centre. Financial Times
Niamh Rowe (2026), US farmers are rejecting multimillion-dollar datacenter bids for their land: ‘I’m not for sale’. The Guardian
Jenna Ruddock (2025). The Real Race for an AI Moratorium: Stopping Data Centers, Tech Policy Press
Ece Yildirim (2026). People Hate AI Even More Than They Hate ICE, Poll Finds. Gizmodo
Lucas Ropek (2026) The public opposition to AI infrastructure is heating up. Techcrunch
