Morning Edition

The Futurist

Sunday, June 21, 2026
AI & Technology Markets & Crypto Ideas Worth Keeping
S&P 500$7,500.58▲ 1.08%
Dow Jones$51,564.70▲ 0.14%
Nasdaq$26,517.93▲ 1.91%
VIX$16.78▲ 2.32%
BITCOIN$63,631.00▲ 2.09%
RIPPLE$1.15▲ 2.10%
ETHEREUM$1,726.84▲ 2.20%
DOGECOIN$0.08▲ 2.26%
01

AI & Technology

404 MEDIA

If AI Is Sentient Then So Is 'Age of Empires II'

Researchers formally demonstrate that standard tests for AI sentience would also flag a 25-year-old strategy game as conscious. The paper's real target: the industry's habit of projecting human-like interiority onto systems that produce human-like outputs.

SIMON WILLISON

GLM-5.2 Is Probably the Most Powerful Text-Only Open Weights LLM

Chinese lab Z.ai dropped a 753B-parameter model under MIT license — open weights, no strings. Willison calls it Opus-level on design tasks, which matters enormously for anyone building outside the closed-model ecosystem.

SALESFORCE / 404 MEDIA

Salesforce's Internal AI Leaderboard Has Teams Competing for Little Trophies

Salesforce built a gamified internal dashboard ranking employees by AI adoption, complete with a "click to see who 👀" shame feature for non-participants. Corporate AI mandates have entered their surveillance era.

COINDESK

Ethereum's Biggest Sandwich Bot Drained of $7.5 Million in Ironic Exploit

The largest MEV sandwich bot on Ethereum — a predator that front-runs user trades — got front-run itself and lost $7.5M. Poetic justice as an on-chain event.

COINDESK

AI Is Making Crypto Security Cheaper, Faster and Harder to Ignore

AI tooling is compressing the cost of smart contract auditing and on-chain threat detection. The same dynamic is also making exploits faster to engineer — a bilateral arms race with lower barriers on both sides.

SIMON WILLISON

Quoting Sean Lynch on MCP's Real Value

Lynch argues MCP's killer feature isn't tool-calling — it's keeping auth flows outside the agent's context window entirely. A narrow insight that reframes the whole protocol debate.

02

X / Twitter Signal

Andrew Ng called out both the US government and Anthropic for actions that demonstrate their power to restrict AI access — referencing the Claude Fable 5 ban. Coming from one of AI's most prominent voices, this is a significant public rebuke of centralized model gatekeeping, and a signal that the open-weights vs. closed-model tension is becoming a political fault line, not just a technical preference.

Ethan Mollick flagged research suggesting AI is commoditizing contract labor by leveling performance across workers. When AI erases the quality gap between a top contractor and an average one, it doesn't just change wages — it restructures why you'd hire anyone at all. This is the quiet economic story underneath the louder benchmark wars.

Mollick also noted that if AI self-improvement is possible even in limited forms, the shipping cadence for both models and AI-native products should accelerate. This is the compounding loop everyone is theorizing about — and he's suggesting early evidence it's already starting.

The IT sector now makes up roughly 38% of the MSCI USA Index. That's not diversification — that's a single thesis wearing an index costume. With Nasdaq up nearly 2% on the day and VIX ticking higher, the concentration risk hiding inside "the market" is worth watching carefully.

03

The Thread

The week's clearest throughline: power over AI access is consolidating fast, and the people building on top of these models are starting to notice. The Claude Fable 5 ban — a government letter prompting Anthropic to restrict its most capable model — spooked enough builders that Greg Isenberg made three separate videos about it. Andrew Ng went public with concern. The open-weights release of GLM-5.2 under MIT license landed in this exact context, and it landed hard.
When the most powerful model on the planet can be switched off by a letter, the architecture of your business becomes a geopolitical question.
The sentience debate got a useful reality check this week. The Age of Empires II paper isn't really about video games — it's about the standards of evidence the industry accepts when it wants to believe something. We anthropomorphize AI because it's convenient, commercially and emotionally. Formalizing that bias is overdue. Meanwhile, Salesforce's trophy leaderboard shows the other face of the same coin: corporations anthropomorphizing AI adoption as a moral virtue, with shame mechanics for the unconverted. On the market side, the sandwich bot getting sandwiched is funny until you realize the same AI tooling that's compressing crypto security costs is also compressing exploit development costs. Mollick's commoditization point applies here too — when the capability gap closes, the edge goes to whoever moves faster. With Nasdaq at all-time highs and IT eating 38% of major indices, faster is the only direction anyone's planning for.