Morning Edition

The Futurist

Thursday, July 02, 2026
AI & Technology Markets & Crypto Ideas Worth Keeping
S&P 500$7,499.36▲ 0.79%
Dow Jones$52,319.20▲ 0.26%
Nasdaq$26,213.72▲ 1.52%
VIX$16.94▲ 2.98%
BITCOIN$58,803.00▼ 0.63%
RIPPLE$1.04▲ 0.30%
ETHEREUM$1,577.63▼ 0.12%
DOGECOIN$0.07▼ 1.22%
01

AI & Technology

404 MEDIA

Scientists Asked AI to Impersonate 112 Public Figures. What Happened Next Is a 'Dire' Warning

Researchers found that people rated AI-generated impersonations of politicians as more authentic, coherent, and relevant than the real thing. The implications for elections and public trust are immediate and severe.

SIMON WILLISON

What's New in Claude Sonnet 5

Claude Sonnet 5 dropped this week alongside the lifting of Commerce Department export controls on Claude Fable 5. Two major Anthropic releases in rapid succession signals the lab is moving fast after a period of regulatory friction.

404 MEDIA

County With 37 Data Centers Asks Schools to 'Conserve Electricity'

Henrico County, Virginia — a major data center hub — is warning residents of a 25% electricity cost spike and telling school workers to close the blinds. The infrastructure cost of AI is landing directly on local taxpayers and kids in classrooms.

404 MEDIA

The AI Tokenpocalypse Is Here

Companies are burning through AI tokens at unsustainable rates while fake AI-generated products flood Etsy, eBay, and Amazon. Two symptoms of the same underlying fever: AI usage is scaling faster than anyone planned for.

404 MEDIA

Apple 'Hide My Email' Vulnerability Reveals Real Email Addresses

A security flaw in Apple's privacy feature may allow attackers to de-anonymize users' hidden email addresses. The researcher who found it says affected users deserve to know it exists.

A16Z

The Case for AI That Improves Itself

Former Google and Anthropic researchers at Mirendil are building AI systems that contribute meaningfully to their own development. Self-accelerating AI is moving from thought experiment to funded startup.

02

X / Twitter Signal

Mollick — who had early access to Claude Fable 5 — warns that without deliberate prompting discipline, the model develops its own "internal bizarre cadence." As the field rushes to publish Fable workflows, he notes how little anyone actually knows about optimal usage patterns for a model this capable. It's a rare note of epistemic humility amid a week of breathless hype.

Karpathy amplified the Etched announcement: $800M raised, $1B+ in customer contracts, and their first racks built after a successful A0 tapeout. A purpose-built AI inference chip company coming out of stealth with those numbers is a signal that the hardware layer of the AI stack is getting very serious, very fast.

New Google research confirms what practitioners already suspect: LLMs hallucinate with high confidence, miss their own knowledge boundaries, and misreport uncertainty — and most calibration fixes are just bolted-on patches. Pair this with the 404 Media story on AI impersonation and the picture gets uncomfortable quickly.

Retail investors now account for just 6% of total Magnificent 7 trading volume over the past five trading days. Whether this is rotation, exhaustion, or quiet exit, retail pulling back from the AI-adjacent mega-caps is a data point worth watching heading into Q3.

03

The Thread

This week's throughline isn't any single model release — it's the collision of AI capability with real-world infrastructure limits. Fable 5 returns from its export-control exile, Claude Sonnet 5 ships, Etched raises $800M to build dedicated inference silicon, and Google drops Nano Banana 2 Lite for speed-sensitive use cases. The model race is genuinely accelerating. But the electricity bill is arriving at the same time.

"A county with 37 data centers is telling teachers to close the blinds to save electricity. That's not a footnote — that's the story."
The deeper tension is between AI's centralized infrastructure costs and its decentralized hype. Henrico County schools absorb the grid strain while Etsy floods with AI-generated fake flowers and retail investors quietly step away from Mag 7. The people building and funding the stack are doing very well. The externalities are landing somewhere else entirely. Meanwhile, the AI-impersonation research deserves more attention than it's getting. When people find synthetic politicians more coherent than real ones, the problem isn't just deepfakes — it's that the bar for "authentic" has already been corroded. Ethan Mollick's caution about Fable developing its own "bizarre cadence" rhymes here: these systems have emergent behavioral signatures that nobody fully controls yet, and we're deploying them into elections, commerce, and identity infrastructure simultaneously.