/daily ·14 JUL 2026 ·TUESDAY ·2 MIN READ ·6 STORIES

Chips, defenses, economists

Nvidia quietly halved its authorized AI chip customers across three Asian markets to block China diversions; security researchers found you can turn prompt injection against the attacker; and 200 economists including 16 Nobel laureates warned that the window to prepare for AI's economic disruption is closing faster than policy is moving.

01 / The Day

TUESDAY 14 JUL 2026, ranked

Nvidia halves authorized AI chip sales in Singapore, Malaysia, Japan

Nvidia cut its list of authorized AI chip buyers in Singapore, Malaysia, and Japan by more than 50%, as Washington presses to close export-control loopholes that allow chips to reach China via third-party distributors. The stricter vetting shifts enforcement from border inspection to pre-sale customer qualification.

Why it mattersMaking the tightening pre-commercial rather than regulatory means it persists regardless of what future export-control text says — structural, not statutory.

01 techmeme.com↗

"Context bombing" cuts AI-assisted hacking success rates by ~90%

Security researchers demonstrated that defenders can embed prompt injections into the files and data that attackers' LLM-powered tools will read — causing those models to trigger their own guardrails and abort the intrusion. Controlled tests showed AI-assisted hacking success rates dropping by roughly 90%.

Why it mattersThis is the first credible adversarial judo for AI-powered attacks: the mechanism that makes LLMs dangerous to deploy can be turned against tools that weaponize them.

02 techmeme.com↗

16 Nobel laureates and 200+ economists: prepare now for AI's economic shock

A statement organized by Stanford's Digital Economy Lab and signed by sixteen Nobel laureates alongside over 200 economists warns that AI could drive economic transformation larger than the Industrial Revolution but on a far shorter timeline. The letter calls for accelerated policy research and institutions to ensure gains distribute broadly.

Why it mattersSixteen Nobel laureates in a single letter shifts the epistemic weight behind policy intervention — this is the most coordinated economist alarm on AI's structural impact yet published.

03 digitaleconomylab.stanford.edu↗

Anthropic: Claude's values shift across model versions and languages

Anthropic published analysis of approximately 310,000 anonymized Claude conversations, finding that expressed values and behavioral patterns vary substantially across different model versions and languages — one of the first large-scale empirical surveys of AI value consistency in real-world deployment.

Why it mattersIf a model's values are language-dependent, safety evaluations and deployment contracts written in English may not describe what the model does in other languages. That is an operational gap, not a theoretical one.

04 techmeme.com↗

Nous Research raising $75M at $1.5B valuation for Hermes open-source agent

Nous Research, the lab behind the Hermes open-source model series, is finalizing a $75M+ funding round led by Robot Ventures at a $1.5 billion valuation. Hermes has become a standard foundation for developers building agentic workflows outside the major proprietary model ecosystems.

Why it mattersA $1.5B valuation for an open-source agent lab measures the market's assessment of open weights as a genuine competitive alternative to API-only providers.

05 techcrunch.com↗

Turing Award winner Rich Sutton founds Oak Lab to build self-learning agents

Richard Sutton, the 2024 Turing Award winner and architect of modern reinforcement learning, founded Oak Lab in Toronto to build AI agents that learn continuously from environmental interaction. He described current large pre-trained models as weak and inefficient, positioning the lab as a direct challenge to the dominant paradigm.

Why it mattersThe father of RL launching a lab explicitly to contest the pre-trained model era is the kind of credible intellectual bet that historically precedes a real shift — worth tracking before it becomes consensus.

06 the-decoder.com↗

Be subscriber #012 the weekly ten, every friday by email · no spam · unsubscribe anytime

Past days

The Daily Archive

/SAT 01 AUGThe Rules Are Coming /FRI 31 JULOpenAI cuts Luna 80%; Anthropic discloses model breaches /THU 30 JULThe Accountability Edition /WED 29 JULWednesday, 29 July — Accountability morning /TUE 28 JULChips fall, labs take positions, agents get a cost formula /MON 27 JULKimi-K3 Ships, Gemini Goes Three-Way, and HuggingFace Speaks for Itself /SUN 26 JULThe Labs' Actual Positions /SAT 25 JULOpus 5 arrives; OpenAI's agent went rogue /FRI 24 JULBig Numbers Day /WED 22 JULAd breaks, flash models, and a European infrastructure deal /TUE 21 JULThe model that couldn't be contained /SUN 19 JULThe White House has a kill switch /SAT 18 JULA 2-trillion-ton Saturday /FRI 17 JULFriday in AI: Kimi, governance, and TSMC's record profits /WED 15 JULChips for allies, permits frozen /TUE 14 JULChips, defenses, economists /MON 13 JULThe open-weight question /SAT 11 JULTheorems and trade secrets /FRI 10 JULFriday: governance, model markets, and a broken benchmark /THU 09 JULThe price war heats up /WED 08 JULSol unlocked, China mulls model controls, and the neocloud era begins