The radical Briefing | Blockchains Finally Found Their Killer App: Crime
New AI browser vulnerabilities exposed, plus: AI vs. college grads, blockchain malware, and croissant-scented stamps
Dear Friend –
If you follow any of the debates between “AI is the best thing since sliced bread” and “AI, and specifically LLMs, are doomed,” it’s getting heated. On one hand, we have the continuous barrage of new releases and updates from major AI companies (did you check out Atlas, OpenAI’s new browser? Anil Dash called it “The Browser That’s Anti-Web”?). On the other hand, you have people like Gary Marcus (see Jeffrey’s pick for his weekly recommended read) and Ed Zitron shouting from the rooftops that AGI is nowhere near the corner, AI companies are unsustainable (on a massive scale), and AI usage is slowing down. Personally, I sit between the two on this one. I largely agree with Ed and Gary, but I also use AI daily (sometimes for hours at a time) and find it genuinely helpful for some tasks. What about you? AI Yay or Nay?
And now, this...
Headlines from the Future
Home cooking has become a dying art form. Latest data from investment firm Apollo shows that more than 50% of U.S. households’ food consumption is away from home. This is good for restaurants and food delivery services, but bad for grocers (and, likely, your waistline). It makes you wonder how this might change with the sharp rise in GLP-1 drug intake (Ozempic and friends)?!
↗ Source
━━━━━
The Battle Royale of GenAI vs. GenZ. The biggest loser in the Gen AI race seem to be GenZ – and more precisely well educated college kids:
The UK tech sector is cutting graduate jobs dramatically – down 46 percent in the past year, with another 53 percent drop projected, according to figures from the Institute of Student Employers (ISE). […] If correct, the survey indicates that AI is starting to close the entry door to tech careers faster than anyone expected. Companies are making short-term efficiency gains at the expense of their long-term talent pipeline, and graduates are seemingly caught in the middle.
↗ Tech industry grad hiring crashes 46% as bots do junior work
━━━━━
The AI’s Are Investing. Someone gave a bunch of frontier LLMs $10,000 real dollars and let them go wild on the investment market (specifically crypto perpetuals on Hyperliquid). It’s fascinating to watch, and so, after a short blip where some of the AIs were trading up, it doesn’t look all that good. We might not quite be at the point where you ditch your financial advisor for ChatGPT.
━━━━━
AI Hallucinations Might Just Be Fine. As much as hallucinations in large language models might not be something we will ever get rid of, they might as well be something which becomes statistically negligible. A new paper shows the way:
“Specifically, we prove that hallucinations can be made statistically negligible, provided that the quality and quantity of the training data are sufficient.”
━━━━━
Using Blockchains to Distribute Immutable Malware. This is getting good – everybody’s 2018 darling technology, the blockchain (remember? “Blockchains will replace the Internet!”) found a new use case: the distribution of malware. And as blockchains come with such lovely features as immutability (what’s recorded on the blockchain can’t be removed), it means you can’t remove the malware once it’s planted.
There’s a wide array of advantages to EtherHiding over more traditional means of delivering malware, which besides bulletproof hosting include leveraging compromised servers.
The decentralization prevents takedowns of the malicious smart contracts because the mechanisms in the blockchains bar the removal of all such contracts.
Similarly, the immutability of the contracts prevents the removal or tampering with the malware by anyone.
Transactions on Ethereum and several other blockchains are effectively anonymous, protecting the hackers’ identities.
Retrieval of malware from the contracts leaves no trace of the access in event logs, providing stealth.
The attackers can update malicious payloads at anytime.
↗ Nation-state hackers deliver malware from “bulletproof” blockchains
━━━━━
Using Images To Trick LLMs. Talking about things getting weird—you can take an image or screenshot that contains invisible (to the human eye) information, upload it to an LLM of your choice (particularly the new breed of AI-powered browsers), and trigger a prompt injection attack. At this point, you ought to be truly careful when using LLMs, especially if you are exposing them to the outside world (e.g., if your business offers an AI-based chatbot).
What we’ve found confirms our initial concerns: indirect prompt injection is not an isolated issue, but a systemic challenge facing the entire category of AI-powered browsers. [...]
As we’ve written before, AI-powered browsers that can take actions on your behalf are powerful yet extremely risky. If you’re signed into sensitive accounts like your bank or your email provider in your browser, simply summarizing a Reddit post could result in an attacker being able to steal money or your private data.
↗ Unseeable prompt injections in screenshots: more vulnerabilities in Comet and other AI browsers
What We Are Reading
🧑🏼🎓 Inside San Francisco’s New AI School: Is This the Future of US Education? Welcome to the most expensive elementary school in San Francisco, where tech billionaires bet that AI can replace homework. @Jane
🤖 Wikipedia: Signs of AI Writing Curious how to detect things written by AI? Wikipedia shares how they do it. @Mafe
🤔 Is Vibe Coding Dying? Stop us if you’ve heard this one before: The vibe coding revolution may not be all it’s been cracked up to be. @Jeffrey
💩 AI-Generated “Workslop” Is Destroying Productivity We’ve been sold the myth that generative AI will supercharge productivity; what’s proliferating instead is ‘workslop’: polished-looking outputs from AI that lack real substance and dump the cognitive burden onto colleagues. @Kacee
💸 They Got to Live a Life of Luxury. Then Came the Fine Print. Buy-Now-Pay-Later services such as Klarna exploded and left behind a wasteland of consumer debt. There is much to unpack here, but just one statistic: two-thirds of all Coachella visitors this year bought their ticket using a BNPL service. And you can get your burrito on installments! @Pascal
Down the Rabbit Hole
🥐 Yum! French post office rolls out croissant-scented stamp.
🥚 The price of eggs compared to the US Consumer Price Index
💥 Gary Larson, the creator of the wonderful The Far Side cartoons is back (somewhat): New Stuff – Fresh from Larson’s Lab
👀 Fascinating insight into the biases in AI systems: AI couldn’t picture a woman like me - until now
🤣 Hey Look, It’s Every AI-Coded Website Ever
🎤 The Rapper 50 Cent adjusted for inflation
🧬 Want to sequence your DNA at home for less than $2k? Yes, you can.
🛝 Ever wondered how the balls in a children’s ball pit are cleaned? Here’s how.
🛏️ Collateral damage: The AWS Outage Bricked People’s $2,700 Smartbeds
Pascal is getting excited for a week of Hawaiian bliss…



I really resonate with what you wrote about the AI debate, it's such a complex space right now! I'm curious how you see the role of regullation evolving in all of this, especially with the 'AI is slowing down' narrative juxtaposed with so many new tools; you always have such a sharp take on these developments.
Re: the AIs are investing...
AIs have been investing since at least the 90s. I've had friends in London who worked at fintech startups (pre-web even) that coded AI systems for making "intelligent" stock market trades.
They all failed, of course. As will the current generation of actors.
The issue today is the same as then: stochastic systems tied up with the anthrocomplexity of public markets do not somehow break the old investment adage of "past performance is not indicative of future results". The so-called smartest guys in the room will still get pantsed by margin calls from snarky Redditors proppping up unsustainable companies just to stick it to the man.
But good luck getting sufficient training data to avoid those out-of-distribution errors.