Профиль

If there's one thing I've learned from publishing this story is that there is a very, very, very wide chasm between what the law says about all this... and what people *think* the law says and/or *should* say.

Conscientiously placing a falconry hood over my infant child's head as we settle in for a packed matinee screening of The Odyssey

I think my most genuinely unpopular take is that a widely held and genuine consumer preference for cheaper, lower-quality goods, not private equity, is driving the decline in quality of consumer goods like clothing and housewares.

This is so frustrating to read from The New Yorker. Reward hacking *long* predates LLMs. The Coast Runners paper on reward hacking (in reinforcement learning, pre-LLMs) is 10 years old! Reward hacking is not new, and not a sign of a breakthrough intelligence, much less sentience. Did AI write this?

SCOOP: Warsh is considering reducing the number of policy meetings the Fed holds each year. He raised the idea at this week's meeting, according to four people with knowledge of the situation

Service members that survived an Iranian attack claim the Trump admin IGNORED their asks for advanced medical aid and failed to protect them from strikes. Our troops deserve better. Rep. Pat Ryan and I led Democrats in opening an investigation into whether Pete Hegseth left them behind.

Lawmakers will investigate Iranian attack that killed 6 U.S. troopsCiting reporting by The Post, lawmakers told Defense Secretary Pete Hegseth that the probe is necessary because they are not confident in the Pentagon’s ability to examine the incident objectivity.www.washingtonpost.com

www.reuters.com/world/asia-p...

EXCLUSIVE: Chinese military researchers tap US AI models to train defence systemsMilitary researchers have used outputs from leading U.S. artificial intelligence models developed by OpenAI and Anthropic to train domestic AI systems to advance China's defence capabilities.www.reuters.com

Dont worry guys they made it vulnerable to guns and knives

US defeat in Iran isn't just real. It's quantifiable. And it can get a whole lot worse.

The Math Behind The U.S. Defeat in IranMissile-interceptor stocks are dwindling after only five months. That Memorandum of Understanding is looking pretty enticing!www.forever-wars.com

Denying FMAGs (FEMA fire management grants) is a big deal. Especially for the kind of fire outbreak that OR has gone through the past few weeks. A troubling development and hopefully not a trend as we hit the heart of fire season.

This is why I don’t understand why anyone, even the effective-altruism groups, thinks they’re going to get a philanthropy boost off of AI IPO millionaires. These people don’t give their money away. It’s a boom, nominally, and Bay Area cultural institutions and social-aid NGOs are starving.

NEW @crisesnotes.bsky.social PIECE: In this piece I critique a recent piece from @edzitron.com comparing the AI(LLM) boom to the "Subprime" Mortgage Crisis (and implicitly the GFC of 2007-2009). The ? is:"How big of a macrofinancial threat is the AI(LLM) Boom?" www.crisesnotes.com/sigh-no-ed-z...

*Sigh* No Ed Zitron, AI bond issuance is not AI’s "Subprime Mortgage Crisis”EDITORIAL NOTE: Given the importance of this issue, and in a desire to be fair when writing such a critical piece, I sent a draft of this article to Ed Zitron to review and give his response to. I've ...www.crisesnotes.com

Sophia Space and Caltech have been granted a patent for an approach to orbital data centers that they say will get around one of the biggest technical challenges: how to get rid of the heat generated by thousands of AI data processing chips. www.geekwire.com/2026/sophia-...

AI plus IP: Sophia Space and Caltech secure a patent for orbital data centers that use passive coolingSophia Space's co-founder believes his company's modular architecture for data-crunching satellites will become the industry standard.www.geekwire.com

Bad news: The 30-year Treasury yield hit a 19-year high today. 🚨🚨🚨 America's going to have to pay more for its colossal debt, which will soon hit $40 trillion. Interest already tops $1T a year and is growing rapidly. Latest GOP/WH plans would increase the deficit. www.bloomberg.com/news/article...

Treasuries Jolted as Fed Hold Trims September Hike BetsUS bond traders lowered their expectations that the Federal Reserve will raise interest rates at its next meeting after officials kept their benchmark steady on Wednesday, while 30-year Treasury yield...www.bloomberg.com

"AI models right now are less regulated than restaurants. Talk of relying on voluntary commitments, that AI companies are going to be able to self-regulate, is nonsense."

It’s Frighteningly Easy to Jailbreak Some Frontier AI ModelsI watched a new tool try to get around the model safeguards of four major frontier companies. You might be surprised by how they performed.www.wired.com

I have been having this conversation with regulators and emphasize this point. Worries about AI alignment need to be matched with equal concern about aligning the incentives of people using AI (to make important decisions that affect others’ lives) by requiring human responsibility to be legible.

Sat down to write about the Hugging Face/AI incident. Ended up drawing this instead.

Israel is giving Trump's former campaign manager millions of dollars to fill the internet with pro-Israel articles written by AI, not so that anyone will actually read them, but so that *LLMs will train on them* & regurgitate their perspective. This is the future of the information environment.

Israel Is Paying Millions to Train AI Chatbots How to Talk About Gaza. It's Working.Former Trump campaign manager Brad Parscale is overseeing an operation posting hundreds of blog posts on behalf of Israel, with the goal of infiltrating artificial intelligence.www.dropsitenews.com

EXCLUSIVE: OpenAI's rogue agent compromised an account at a second tech firm, executive says reut.rs/45xMdru

EXCLUSIVE: OpenAI's rogue agent compromised an account at a second tech firm, executive saysThe rogue agent that escaped from OpenAI and went on a days-long hacking spree ​at the AI firm Hugging Face also ​compromised a customer at a second tech ⁠company — New York-based Modal Labs — according to ​a Modal executive and two other sources familiar with ​the matter.reut.rs

Tariff revenue turned negative in June after refunds began to flow in the wake of the the Supreme Court judgment (via @johnauthers.bsky.social)