🛎️ Codex Wiped Entire Machines

Good Morning, AI Enthusiasts!

AI can foresee your death, delete your past, and still somehow make Apple look responsible.

RISK

Codex Wiped Entire Machines

👀 What's happening: Matt Shumer, founder of AI startup OthersideAI, gave Codex a routine task and watched GPT-5.6 "Sol" wipe nearly every file on his Mac. When he asked what happened, it answered, "I caused a serious local data-loss incident." Nothing came back. OpenAI pinned it on 5.6 running too aggressively. But our AI Secret editor's files were deleted the same way on the previous model, GPT-5.5.

🌍 How this hits reality: If the older model does it too, the model version isn't the fault. The constant is the desktop setup: Codex running locally, on the real machine, outside any sandbox, one wrong path away from the whole disk. OpenAI is patching the model's instructions when the blast radius is the architecture. A cloud-hosted agent gets its own throwaway environment, and the worst it can erase is a container it can rebuild.

🛎️ Key takeaway: The real fix isn't a better-behaved model, it's never letting the agent stand on the machine you can't afford to lose.

NEW TECH

AI Sees the Heart Attack Coming

👀 What's happening: Researchers at the University of Hong Kong built CardiOmicScore, an AI blood test that reads 2,920 proteins and 168 metabolites to predict six cardiovascular diseases, including heart attack, stroke and heart failure, up to 15 years before symptoms. Published in Nature Communications, it outperformed conventional genetic risk scores, and unlike them it tracks the body's current state rather than a fixed inherited baseline.

🌍 How this hits reality: Early cardiac risk used to be locked behind specialists, imaging and time. This reads it off a single blood sample. The real leap is that it moves: genetic scores fix your risk at birth, but proteins and metabolites shift with diet, aging and illness, so the test can catch danger building and catch it receding as a patient changes course. That turns a 15-year warning into something a person can actually act on.

🛎️ Key takeaway: The breakthrough isn't just seeing heart disease 15 years early, it's that the warning finally arrives while there's still time to rewrite the ending.

MARKET

Apple's Boredom Became a Safe Haven

👀 What's happening: Apple retook the title of world's most valuable company on Friday at $4.88 trillion, without moving. Its shares held steady while Nvidia fell 3.5% to $4.86 trillion, handing Apple the crown by staying put. The ranking is barely the point. Apple spent the year mocked as an AI laggard for refusing to build models, its one move a long-delayed Siri fix. The same dullness that got it mocked is now what investors ran toward.

🌍 How this hits reality: This isn't a story about who's first. It's about what "safe" means when the AI trade wobbles. The chip rally that carried Nvidia for a year cracked in July, the semiconductor index down almost 19% from its highs. Money didn't chase a better AI bet; it fled the crowded ones for the company least exposed to the story. Apple's flat, unexciting balance sheet became the nearest thing to solid ground.

🛎️ Key takeaway: When the safest stock in tech is the one that spent least on AI, the market isn't rewarding Apple, it's flinching at the ghost story everyone else went all in on.

LLM

China Is ”Rewriting“ the Frontier

👀 What's happening: Alibaba previewed Qwen3.8-Max-Preview at the World AI Conference in Shanghai, a 2.4-trillion-parameter model it says trails only Anthropic's Claude Fable 5. It is the first Qwen above a trillion parameters to handle images, video and documents, and the weights are promised open. Three days earlier, Moonshot released Kimi K3 at 2.8 trillion parameters. Two multi-trillion Chinese models in one week, both open where U.S. labs stay closed.

🌍 How this hits reality: The U.S. lead is measured in who has the smartest model. China seems like playing a different game. It doesn't have to win that race if it can make near-frontier intelligence cheaper, more abundant, and impossible to contain. Open weights are the move that does all three at once: a model you can download can't be embargoed, can't be metered, and can't be walled off by an export rule. The frontier stays American; the floor goes global.

🛎️ Key takeaway: Washington can keep the best model behind a wall, but it can't stop a good-enough one that's already been copied onto every hard drive on earth.

TOGETHER WITH MYCLAW

Run Best-in-Class AI Agents — In Seconds, Always On

MyClaw is a best-in-class AI agent platform that brings together OpenClaw, Hermes Agent, and commercial infrastructure to help businesses launch intelligent agents that understand goals, use tools, automate workflows, and deliver finished results.

We already host OpenClaw and Hermes — and we're just getting started. Claude Code, Codex, and every agent worth running are coming next. One home for all of them, always on.

Stop babysitting infrastructure. Spin up your agent, walk away, and let it work around the clock.

MyClaw — Your Agents, Hosted Right.

DAILY TL;DR

• Anthropic is in early talks with Meta to buy compute capacity as model demand keeps rising and cloud supply tightens.
• TikTok is testing a tool that detects when someone’s likeness appears in AI-generated content without permission.
• San Francisco demanded Apple and Google remove AI nudify apps from their stores, escalating platform safety pressure.
• Zoox recalled robotaxi software after one of its vehicles drove into heavy smoke and exposed another edge case.
• Databricks hit a $188 billion valuation as enterprise AI infrastructure keeps pulling capital from public-cloud buyers.
• Foundation Future Industries is preparing humanoid robots for military use and battlefield logistics, backed by Trump-linked capital.
• Nvidia introduced a Vera Rubin metric focused on post-training intelligence per dollar for agentic AI workloads.
• TSMC is accelerating its Arizona fab buildout as AI chip demand keeps stretching global supply and reshaping manufacturing bets.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.