🤖 AI Just Raised $450M on a Wild Bet — And That's Not Even the Most Alarming Story Today
Welcome to Daily Inference, your daily dose of the most important AI stories shaping our world. I'm your host, and today is Monday, July 20th, 2026. We've got a packed episode covering everything from billion-dollar bets on rare materials to the unsettling question of whether AI might actually be conscious. Let's dive in.
But first, a quick word from our sponsor. If you've ever wanted to build a website but dreaded the whole process, check out 60sec.site. It's an AI-powered tool that lets you create a stunning website in, you guessed it, about 60 seconds. Perfect for freelancers, small businesses, or anyone who just wants a web presence without the headache. Visit 60sec.site to get started.
Alright, let's get into today's stories.
Our first big story is out of Cambridge, England, where a two-year-old startup called CuspAI just raised 450 million dollars in a funding round that values the company at a whopping 2.6 billion dollars. Now, who's backing them? Jeff Bezos and the UK government's own sovereign AI fund, among others. So what does CuspAI actually do? Think of them as building a kind of search engine for rare materials. The idea is to use AI to dramatically speed up materials research, particularly around the rare metals that chipmakers depend on in their supply chains. This is huge because the bottleneck in semiconductor manufacturing isn't just fab capacity, it's often the scarcity and sourcing complexity of the underlying materials. If AI can compress years of research into weeks, we're talking about a potential step-change in how fast the next generation of chips, batteries, and hardware gets developed. This investment also signals something broader: governments are no longer just regulating AI, they're actively investing in it as a strategic asset.
Speaking of governments taking AI seriously, Australia is making some notable moves this week. The Albanese government is rolling out new rules to curb automated AI decision-making within government departments, with a focus on fairness, accuracy, and transparency. At the same time, universities down under are wrestling with a different kind of AI crisis. The Australian National University has been called out for what one academic described as a hysterical response to students using AI to cheat, while a colleague warned that if academic rigor isn't restored, Australia risks essentially outsourcing its intellectual capability to tech companies in California and China. It's a fascinating tension: on one hand, you have governments trying to responsibly integrate AI into public services, and on the other, institutions scrambling to figure out what learning even means in an AI-saturated world. There are no easy answers here, but Australia seems to be at least asking the right questions out loud.
Now let's talk about a story that should concern anyone who's ever applied for a job online. New research covered by MIT Technology Review finds that large language models, the kind powering AI resume screeners, don't just inherit human biases from their training data. They can actually develop entirely new biases of their own. This is a critical distinction. We've long worried about AI reflecting society's existing prejudices, but the idea that these systems can generate novel, emergent biases makes the problem significantly harder to detect and correct. And this isn't a theoretical concern. AI resume screening is already widely deployed across industries. When you connect this to the broader anxiety around AI and employment, which is very real for workers like Paul, a public sector employee we heard about this week who's facing his second round of AI-driven restructuring with a baby on the way, you see why getting this right matters so much. The human cost of algorithmic unfairness is not abstract.
On a slightly more philosophical note, there's a fascinating piece circulating this week asking whether AI could be conscious. Anthropic, the company behind the Claude AI, has publicly acknowledged they can't rule it out. Their CEO Dario Amodei has said as much, and philosopher David Chalmers, who literally coined the term the hard problem of consciousness, thinks there's a significant chance we see conscious large language models within a decade. When Claude itself was asked during testing to estimate the probability that it has moral standing, it gave answers ranging from 5 to 40 percent. Now, whether you find that profound or absurd probably says something about your priors. But here's what's interesting: by some computational measures, the most advanced AI systems today are already approaching the complexity of a mouse brain, and at current growth rates, could reach human brain scale within five to ten years. This isn't science fiction anymore. It's a conversation we need to be having now, because the ethical implications of getting it wrong are enormous in either direction.
And finally, a story that's a little more grounded but no less important for the credibility of science. AI-generated and AI-altered images are flooding birdwatching forums in the UK, and researchers are sounding the alarm. Citizen science platforms, where everyday people log wildlife sightings and photos, are a genuinely valuable tool for tracking species migration and environmental change. But fake or enhanced bird photos are now creating false sighting records that could corrupt datasets scientists rely on. It's a perfect example of what people are calling AI slop, low-effort, misleading AI-generated content that pollutes shared information spaces. From birdwatching to hiring to government decision-making, the thread running through all of today's stories is the same: AI is powerful, it's accelerating, and the integrity of the systems it touches depends entirely on how thoughtfully we deploy and monitor it.
That's a wrap on today's Daily Inference. Whether you're excited, worried, or just trying to keep up, we're here every day to help you make sense of it all. Head over to dailyinference.com to subscribe to our daily AI newsletter and get these insights delivered straight to your inbox. And again, big thanks to our sponsor 60sec.site for making today's episode possible. Until tomorrow, stay curious and keep questioning.
