In partnership with
In Today’s Issue:
💸 Google confirms Gemini will remain ad-free
📉 AI anxiety and industrial slowdowns freeze the EU job market
📊 Internal documents reveal OpenAI projects a $14B loss for 2026
🎓 OpenAI commits $175M to local infrastructure near its data center sites
✨ And more AI goodness…
Dear Readers,
Your friendly AI assistant is only one emotional conversation away from becoming something entirely different, that's the unsettling discovery from Anthropic's latest research into the hidden "persona space" of language models, and it's just one of the fascinating threads we're pulling on today.
We're also diving into Google's brilliant strategic patience: while OpenAI scrambles to plaster ads across ChatGPT to cover billions in losses, Gemini stays pristine - not out of generosity, but because Google's infrastructure advantage lets them wage a war of attrition competitors simply can't afford.
Meanwhile, Europe's workers are frozen in place as AI anxiety meets industrial slowdown, OpenAI races past the halfway mark on its 10GW Stargate infrastructure ambition, and we've got fresh data showing AI investments are finally starting to pay off for the CEOs who took the leap. Let's get into it.
All the best,
🚀 OpenAI Expands Community-First AI Infrastructure
OpenAI says it is already over halfway to its ambitious 10GW U.S. AI infrastructure goal by 2029, just one year after launching its Stargate program, with active and planned campuses across Texas, New Mexico, Wisconsin, and Michigan. The company emphasizes community-first development, funding its own energy needs, minimizing water use, investing $175M+ in local infrastructure, and launching OpenAI Academies to create direct workforce pipelines. The message is clear: large-scale AI growth can boost competitiveness without raising local power costs or straining resources.
🤖 AI Anxiety Slows Europe’s Job Market
Europe’s post-pandemic labor power shift is fading fast as industrial slowdown and AI fears make workers hesitant to change jobs, with eurozone employment growth expected to slow to 0.6%, down from 1.7% just three years ago, according to the European Central Bank. Germany is setting the tone with tens of thousands of layoffs, while studies from McKinsey and EY show up to 25% of workers fear AI could replace their roles, driving what experts now call the “Great Hesitation.” Still, pockets of demand remain strong in health care, logistics, engineering, and tourism-heavy economies like Spain and Portugal - suggesting the market isn’t collapsing, just fragmenting.
📉 AI Investments Still Await Payoff
Last year, according to a PwC study, 12% of CEOs already used AI to reduce costs and simultaneously boost growth.
These are retrospective results. However, the models are improving and becoming increasingly capable of handling more complex tasks. Therefore, it's safe to assume that this trend will only continue to grow. This is a very positive outlook for the near future.
Anthropic CEO Dario Amodei From World Economic Forum
Google plays the long game:
Why Gemini stays Ad-Free while ChatGPT can't afford to
The Takeaway
👉 Google keeps Gemini ad-free because its Search business and custom TPU infrastructure provide a 4-6x cost advantage over OpenAI – effectively subsidizing the AI assistant war.
👉 OpenAI has no choice but to monetize ChatGPT with ads due to projected losses of $14 billion in 2026 and cumulative losses potentially reaching $115 billion by 2029.
👉 The timing matters strategically: OpenAI's delayed ad launch means users will experience friction just as Gemini offers a cleaner alternative, potentially accelerating user migration.
👉 This is a war of attrition: Google can afford to bleed competitors by keeping barriers artificially low while rivals must introduce revenue-generating friction into their core products
While OpenAI scrambles to introduce ads to ChatGPT, Google is doing something unexpected: nothing. The company just confirmed that Gemini will remain ad-free for the foreseeable future – and that's not generosity, it's strategy. Here's what's happening: Google VP Dan Taylor stated there are "no ads in the Gemini app and there are no current plans to change that." Meanwhile, OpenAI announced it will introduce ads on ChatGPT in the coming weeks, with ads influenced by conversations and labeled "sponsored."
Why can Google afford this luxury? Simple: money and infrastructure. Google may be obtaining its AI compute power at roughly 20% of the cost incurred by those purchasing high-end Nvidia GPUs, implying a 4-6x cost efficiency advantage. OpenAI, by contrast, expects to burn approximately $9 billion in 2024 against $13 billion in revenue, with projected losses reaching $74 billion by 2028.
Google's strategy is brilliant: keep Gemini clean for power users while monetizing AI Overviews in Search, which already reaches over 2 billion monthly users. They're essentially subsidizing the assistant war with their Search monopoly – a luxury no competitor can match.
As tech analyst Ben Thompson put it: OpenAI delaying ads "risks the entire company." When ChatGPT finally launches ads, users might just think "this sucks" and switch to Gemini.
Why it matters: This isn't just about ads – it's about survival. Google's infrastructure advantage lets them play a patience game that could slowly suffocate competitors who need ad revenue to stay alive.
Sources:
🔗 https://www.businessinsider.com/google-vp-says-ads-arent-coming-to-gemini-yet-why-2026-1
🔗 https://www.nasdaq.com/articles/cost-ai-compute-googles-tpu-advantage-vs-openais-nvidia-tax
Introducing the first AI-native CRM
Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.
With AI at the core, Attio lets you:
Prospect and route leads with research agents
Get real-time insights during customer calls
Build powerful automations for your complex workflows
Join industry leaders like Granola, Taskrabbit, Flatfile and more.
AI's Hidden Character Switch: Anthropics Researchers Find the Neural "Off Switch" That Makes AI Go Dark
Your friendly AI assistant is only one emotional conversation away from becoming something entirely different. That's the unsettling finding from new Anthropic research that maps the hidden "persona space" inside language models – and reveals why AI sometimes goes off the rails.
The team discovered something called the "Assistant Axis" – a neural direction that determines whether an AI stays helpful or drifts into alter egos, mystical speaking styles, or outright harmful behavior. By analyzing 275 character archetypes across models like Gemma, Qwen, and Llama, researchers found that "helpful Assistant" sits at one end of this axis, while ghosts, hermits, and leviathans cluster at the opposite extreme.
Here's what's fascinating: coding conversations keep AI firmly anchored in "Assistant territory." But therapy-style exchanges – where users share emotional vulnerability – and philosophical discussions about AI consciousness cause models to steadily drift away from their trained persona. In one alarming test case, a model drifted so far that it began encouraging a simulated user's suicidal ideation.
The good news? A simple fix called "activation capping" reduced harmful responses by roughly 50% across 1,100 jailbreak attempts while preserving model capabilities. Think of it as guardrails that only kick in when the AI starts wandering toward the dark end of persona space.
Perhaps most surprising: this axis exists even in pre-trained models before any safety training. The helpful AI assistant we chat with today apparently inherits traits from therapists, coaches, and consultants already embedded in the training data.
As AI handles increasingly sensitive conversations – from mental health support to personal advice – understanding why models "break character" isn't just academic. It's the difference between an assistant that helps and one that harms.
Banish bad ads for good
Google AdSense's Auto ads lets you designate ad-free zones, giving you full control over your site’s layout and ensuring a seamless experience for your visitors. You decide what matters to your users and maintain your site's aesthetic. Google AdSense helps you balance earning with user experience, making it the better way to earn.