How to breathe air when we’re drowning in big tech’s slop

June 21, 2026 by Ketan Joshi In Uncategorized 3 Comments

There’s a chance you saw this story; most likely on your social media feed. Jeff Bezos, head of Amazon and builder of many thirsty data centres, purportedly said this:

The full quote included in the article at The Print, a large Indian media outlet, is rage-inducing:

“Biological limits are real, but digital potential is infinite. If we starve our data infrastructure of cooling resources just to sustain baseline human comfort, we are actively delaying the birth of a super-intelligence that could solve all of our resource problems in the first place. Sometimes you have to prioritise the intelligence that will save us over the biology that slows us down”

The only reason I immediately clocked this as fake is because Bezos has real greenwashing skills. Whether it’s on-stage conversation or through his massive philanthropic fund, he is a master of appearing environmentally virtuous to exact people he needs to appear virtuous to. Musk, Altman and Zuckerberg are dull human chatbots blurting out the last thing they heard. Bezos does bad things in a more subtle way.

If it were something I didn’t know so well, I 100% would have shared it myself, because it was published in a major media outlet.

At no point during the full 50 minute interview cited by the article does Bezos say the words about water, and there’s no record of him ever having said them. It is weird: the article is so confident. A keen-eyed Bluesky user noticed that the fabricated quote seems to match a parody post by a satirical¹ Indian news account from the day before:

View this post on Instagram

There is a decent chance what happened here is the journalist involved queried a generative AI chatbot about what Bezos said at the event, and that chatbot drew on the satirical post above and regurgitated it as an accurate statement (amusingly, The Print’s article was first shared on Reddit in the ‘Not the Onion’ subreddit).

There is precedent for this: if you remember Google’s AI overviews telling people to eat rocks, there’s a good chance that response drew on a satirical article at The Onion. More recently, scientists invented a fake disease and found chatbots immediately reproduced it and treated it as real. Days ago, it was revealed Google’s AI overview was insisting that a fictional horror creature is in fact a living organism.

I emailed the media outlet to ask if a chatbot was involved in the production of the article, and didn’t receive a reply before publication of this post. But given the precedent, and the sheer weirdness of copy-pasting a parody instagram post, I’m going to say it’s close to certain.

The live article has now been corrected, with a vague note at the bottom. But it broke containment almost immediately.

“The Verified Post”, a site which bills itself with the tagline “Verified. Researched. Told” published a blatantly AI-generated long-form article reproducing the fabrication. The New Indian Express (also a pretty large site) published a counter, not pointing out that Bezos’ remarks were fabricated, but that they were simply taken out of context based on “a closer analysis of Bezos’s comments”. This, too, is blatantly AI-generated.

Naturally, there are thousands of social media posts across many networks with engagement in the tens of thousands per post, each repeating the fabricated quote.

It has even already manifested in chatbot responses. Claude, Ecosia AI and Mistral all just breezily repeated the claim. Gemini seemed to ingest a source pointing out that it was fake, but blended its response with the AI-generated New Indian Express post that said the issue was simply lacking ‘context’. Remarkably, the only chatbot that pointed out it was fabricated was OpenAI’s ChatGPT.

We’re all breathing slop now

From the moment of origination, to its presence in a trusted media outlet, to its spread across social media, chatbot output and even other trusted media outlets, this is a lie that was strapped to the jet engine of AI-powered bullshit. Every lie travels at the speed of light, now.

This isn’t a new thing for media outlets. Thanks to either individual journalist or contributor usage or enforced institutional reliance on generative AI, there is an ever-growing list of colossal fuckups, which you can peruse in full:

As the great Abebe Birhane wrote in a recent paper, the direct capture of media outlets is a major component in how big tech wields its power:

“In an analysis of 1, 000 media articles “[c]orporate actors [were] by far the most frequently quoted by journalists covering AI, with no civil society voices featuring amongst the top-25 most-quoted people or organisations”………Big Tech also exerts its influence over media through ownership, funding of journalism, shaping state media regulation, targeting media (policy) research institutions and universities, and by providing platforms through which the media comes to reach their audiences. In doing so, Big Tech has created an infrastructure in which media and journalists help reproduce, disseminate, and normalize industry narratives, or face exclusion”

This is bad enough. But this is corruption on a scale that I previously thought was impossible. It is a total collapse of the basic drive towards reality in an institution that purportedly offers that as its main selling point.

We are in a weird space right now, where a shockingly severe lie (like a fabricated quote) is paired with the cognitive shortcuts we have in place, like “I trust this media outlet”. Everyone who shared the original Print article had no reason not to: it’s a huge outlet, with stated editorial policies.

Thanks to the failure of these institutions to reject a software system that produces powerfully convincing fabrications, journalism as an industry is seeing the consequences play out. And those consequences will erode trust, killing that mental shortcut we used to lighten our cognitive load when strolling through the world of digital information. Now, we have to carry the infuriating weight of having to check everything, including stuff from the industry that once offered pre-checked information.

The climate movement and generative AI

Apart from the basic moral issue that we shouldn’t believe and spread things that aren’t true, a fabricated Bezos quote used to criticise tech also feeds the ongoing meme that concerns around the industry’s water impacts are “fake”; originated from effective altruist bloggers and spread by OpenAI’s CEO Sam Altman. They are, in fact, very real.

High-polluting industries benefit directly and obviously from the collapse of trust, shared reality and information quality. It is why the fossil fuel industry has spent billions over decades on not just disinformation campaigns but targeting those most trusted (like scientists or journalists) with legal attacks and smear campaigns. Big tech wants the same outcomes, but instead of external attacks, they’re setting fire to those institutions from the inside out.

Whenever I do a good job of debunking climate misinfo these days, AI bots accuse me of being an AI-generated character in order to convince the public what I’m saying isn’t true 🤦‍♀️

— Alaina Wood (@thegarbagequeen.bsky.social) 2026-06-20T15:21:56.452Z

Where this gets really scary is the spread of reliance on generative AI not just in media or science, but in spaces specifically meant to be pushing back on fossil fuels, advocating for climate solutions and criticising big tech.

I frequently see AI-generated content being used in opposition to data centre infrastructure, big tech and AI itself. Google’s AI overviews are regularly being cited as a source for tech-critical information. As I mentioned the other day, climate organisations are increasingly signing up to chatbot subscriptions. There are dedicated climate solutions accounts that are 100% AI generated.

A recent USC Annenberg survey among climate communicators found the majority are already relying on big tech’s inaccurate text generation systems for their work as trusted sources of climate information. When asked “How frequently do you use Al for climate communication work?”, three quarters use it some amount and about 40% use it at least once a week, or daily. A quarter never use it all (it was 30% in 2024).

This hits hard for me. I don’t think there has ever been a moment where large climate groups have ever been so directly involved in the spread of massive new polluting fossil fuel infrastructure. I mean this literally: the data show Google and Anthropic comprise a third of climate communicator use and both of these companies lease compute capacity from the racist, community-choking ultra-polluting methane gas plant in Memphis. I could churn out 2,000 words right now about the various fossil gas plants being built by the other companies in the list.

Anthropic just announced a dedicated army of 1,000 AI-psychosis-ridden missionaries tasked with converting nonprofits into high-polluting slop reliance (something they frame as ‘philanthropy’). Are we dealing with what it means for there to be an embedded fossil fuelled software proselytisation force within nonprofits? I don’t think we are.

Climate action is increasing its reliance on completely avoidable new fossil fuel infrastructure, and what it gets in return is more lies, less trust and direct participation in big tech’s attempt to destroy access to reality. This is a bad deal, and a moment of total moral collapse, if it keeps getting worse.

Breathing air

There is something important in the second question in the climate communicator survey. Those with the strongest feelings are opposed to genAI, and those with the weakest feelings are supportive.

What this tells me is that any company-wide policy to at least clamp strict controls on use, if not outright refuse its formal acceptance, wouldn’t face strong opposition. I would bet that survey results for journalists would be roughly the same.

Yes: climate activists have been concern-trolled for decades because they drove a fossil-fuelled car to a climate protest. But generative AI is a completely new technology, not a fuel baked into society since before we were born. It barely existed just a few years ago. GenAI is at the start of becoming unavoidable.

There is no excuse for media outlets or highly-trusted institutions like NGOs, academia or expert professions marbling the most powerful engine of convincing lies throughout the fundamental production of their work. As you can see from this story, all it does is take a sledgehammer to the foundations of shared reality.

The only way we find air to breathe when we’re drowning is slop is placing trust in those who stay wedded to real research and writing work. At the bare minimum, media outlets should be using their outright refusal of generative AI as a key selling point. Generative AI has no internal, designed momentum towards truth and accuracy (beyond the absurdly diminishing returns of energy-hungry multi-layered LLMs), but a person or an institution can (and should).

There needs to be a return of a desire to be as right as we can, admit when we’re wrong and drop real consequences on the people who breach trust by taking the dirty deal of convenience through accuracy sacrifice.

Being pro-correctness, anti-theft and anti-fossil fuels automatically places you in an aggressively anti-generative-AI position. Yet I feel millions of people who fit those criteria haven’t realised they’re against generative AI, yet. The more people realise they were always against this, the more air we find to breathe.

1. This is pretty shitty satire. It’s not funny enough to be immediately identifiable as a joke, and I think many of these accounts are basically just publishing fabrications for clicks, trading on the misunderstanding of most of their audience. It’s gross and I hate it, and I hate the people behind it ↩︎