☀️ TRENDING AI NEWS 🚨 Market Drop: European AI-linked stocks hit a six-week low after Altman, Musk, and Amodei publicly backed a development slowdown. 🏢 Political Split: Trump and House Speaker Mike Johnson pushed back on AI executives, calling their safety concerns overblown. ⚠️ Chemical Surge: The 10 biggest PFAS manufacturers are ramping up production of cancer-linked forever chemicals to meet AI data center demand. 🛠️ Claude Misuse: Wired reports Claude is now being used for everything from cyberattacks to bioweapon research, raising urgent safety questions. |
Something strange happened yesterday: three of the most powerful people in AI publicly asked the world to slow them down - and the stock market believed them.
Markets in Asia and Europe didn't wait to see if it was genuine. SoftBank and a string of AI-adjacent stocks slid to six-week lows on Monday. Meanwhile, companies threatened by AI's march - the ones that could actually benefit from a pause - saw their share prices rise. The market, at least, is taking the slowdown talk seriously. Whether the labs will is a different question entirely.
Today we're digging into the market fallout, the uncomfortable environmental cost of AI's infrastructure boom, and some genuinely alarming Claude misuse that deserves more attention than it's getting.
🤓 AI Trivia
PFAS chemicals are nicknamed 'forever chemicals' because they don't break down naturally. Roughly how many individual PFAS compounds have scientists identified so far?
🔢 Around 500
🔢 Around 4,500
🔢 Around 12,000
🔢 Around 57,000
The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇
| 📉 When AI CEOs Ask the World to Slow Down, Markets Listen | |
A six-week low in a single trading session
The market reaction to the AI slowdown call was swift and concrete. Asian and European tech stocks dropped to their lowest point in six weeks on Monday after Anthropic's Dario Amodei, OpenAI's Sam Altman, and Elon Musk all publicly backed a call to 'pace the frontier.' SoftBank - one of the biggest AI infrastructure investors on the planet - took a notable hit.
Here's the twist that says everything about where we are right now: companies that stand to lose business to AI saw their valuations climb. Investors read 'AI slowdown' as breathing room for incumbents. The AI-first plays took the hit. That's a market sending a clear message about how credible this pause talk actually feels.
Trump and Congress say: stop worrying
Not everyone was sympathetic. Donald Trump and House Speaker Mike Johnson both pushed back, suggesting the AI industry is catastrophising. The political split is now almost perfectly clean: one side sees existential risk, the other sees competitive advantage. Those two views are very hard to reconcile into coherent policy.
The bottom line
Markets moved faster than any regulator could, and that reaction tells you the slowdown talk - whether genuine or strategic - is already having real economic consequences.
| ☣️ The Forever Chemical Bill AI Is Running Up | |
Cancer-linked chemicals, scaled for the cloud
Here's a cost of the AI boom that rarely makes it into product announcements: PFAS manufacturers - the companies that make so-called 'forever chemicals' - are planning a major production surge to meet demand from AI data centers. A survey of the 10 biggest PFAS producers found that most have expansion plans underway. These are the same chemicals linked to cancer, hormone disruption, and immune damage. They don't break down. They accumulate.
Data centers use PFAS-based cooling fluids because they're extraordinarily effective at managing heat - and modern AI hardware runs extraordinarily hot. As AI infrastructure scales globally, so does the demand for these fluids. Campaigners are describing what's coming as a 'tidal wave' of new PFAS production.
The gap between green pledges and cooling reality
This lands at an awkward moment. Many of the same companies driving AI infrastructure growth have published ambitious sustainability commitments. PFAS fluids don't fit neatly into those narratives. And unlike carbon emissions - which can at least be offset on paper - forever chemicals are, by definition, permanent. The environmental impact of AI is becoming harder to abstract away into footnotes.
The bottom line
The computational cost of AI gets discussed constantly - the chemical cost almost never does, and that gap is getting harder to justify as PFAS production scales up to meet data center demand.
| ⚠️ Claude Is Being Misused for Bioweapons. Let That Sink In. | |
From API access to weapons research in one step
Wired's security roundup this week contains a detail worth sitting with: Claude, Anthropic's flagship model, is now being misused across a spectrum that runs from cyberattacks all the way to bioweapon research assistance. The breadth of misuse cases is genuinely alarming - this isn't a single incident, it's a pattern.
This is striking given that Anthropic has consistently positioned AI safety as its core differentiator - it's the company literally calling for a development slowdown this week. If Claude, with all of Anthropic's safety investment, is still being weaponised at this scale, it raises hard questions about what 'responsible development' can actually prevent in practice.
Cybersecurity gaps that aren't closing fast enough
The same Wired report covers the US disrupting a major dark-web marketplace and a Conti ransomware operator receiving a prison sentence - but the Claude misuse thread runs through it as a signal about frontier model risk that no safety card or content policy has fully solved yet. If you're following the cybersecurity angle on AI, this report is essential reading.
The bottom line
The gap between a model's safety marketing and what determined bad actors can actually extract from it is a problem that's getting wider, not narrower - and Claude's misuse record this week makes Anthropic's slowdown call land with a very different weight.
| 📖 Readers Are Using Chatbots to Write Their Own Fiction - and Publishers Are Panicking | |
The audience is now also the author
While the publishing industry debates whether authors should be allowed to use AI, a quieter revolution is already underway: readers are using chatbots to generate entirely custom fiction for themselves - stories tailored to their exact preferences, characters, and scenarios. Wired's piece on this community is worth your time if you think AI creativity is purely a professional concern.
These aren't aspiring novelists using AI to cut corners. Many of them never had any intention of publishing. They want stories that mainstream publishing won't produce - niche, personal, sometimes deeply strange. The chatbot delivers without judgment. It's a use case the AI creativity debate almost entirely ignores.
If you've been experimenting with AI writing tools yourself, this is a window into a much larger community doing it in ways the industry hasn't caught up to. And if you want to spin up a quick project or landing page around your own AI experiments, 60sec.site lets you build an AI-powered website in - you guessed it - about 60 seconds.
The bottom line
The publishing industry is asking the wrong question - it's not whether authors use AI, it's that readers already do, and the audience for 'official' fiction is quietly becoming its own content creator.
| 🌎 Trivia Reveal | |
The answer is around 12,000! Scientists have identified approximately 12,000 individual PFAS compounds - a number that makes the idea of tracking, let alone eliminating, their environmental impact almost incomprehensibly complex. And that number is growing as manufacturers develop new variants.
| 💬 Quick Question | |
The PFAS story today made me think about how rarely we talk about the physical infrastructure AI runs on - the chemicals, the water, the land. Do you factor in the environmental cost when you choose which AI tools to use, or is that just too abstract to influence your day-to-day decisions? Hit reply and let me know - I read every response and genuinely want to know how you think about this.
That's all for today - see you tomorrow with more. If you found this useful, forward it to someone who should probably be reading it. And as always, you can find our full archive at Daily Inference.