Zendoric

Zendoric

AI analysis and perspective, every day.

Updated: September 3, 2026

🔥 In focus today

Three seconds of voice is all it takes, and AI fraud's safeguards are built to investigate, not prevent

🕒 Published on Zendoric: September 3, 2026 · 10:20

Three seconds of recorded audio is now enough to clone a voice well enough to fool a parent, and the FBI has started counting the damage: more than $893 million lost to AI-enabled crime in 2025, $352 million of it from people over 60. The tools' security features are real, but almost all of them work after the money is gone. That is the whole problem.

Read the analysis🎬 Video

Anthropic's $40B compute lease from Musk turns a safety promise into a rival's contract clause

🕒 Published on Zendoric: September 3, 2026 · 10:20

Anthropic is reportedly paying xAI/SpaceX about $1.25 billion a month through May 2029 — over $40 billion — for exclusive use of the 220,000-GPU Colossus 1 site in Memphis. The clause drawing attention would let SpaceX reclaim that compute if Claude is found to harm humanity. Neither company has jointly confirmed the wording, and that ambiguity is the story.

Read the analysis🎬 Video

Visa cuts 2,600 jobs and calls it efficiency: the AI part is inference, not a quote

🕒 Published on Zendoric: September 3, 2026 · 10:20

Visa is laying off about 2,600 people, roughly 7% of its 34,000-strong workforce, mostly in product and technology. The CEO memo confirmed to Fast Company talks about efficiency and reinvestment, not about AI. That gap between what was said and how it is being read is the most useful thing in this story.

Read the analysis🎬 Video

ChatGPT, Claude, Gemini and Grok Went Dark at Once: AI Now Has Infrastructure Status Without Infrastructure Rules

🕒 Published on Zendoric: September 3, 2026 · 10:20

Every major chatbot failed at roughly the same moment on Thursday morning, and nobody has said why. The jokes wrote themselves — "Finally I can see the Sun!" — but the real story is that a technology millions now depend on has no disclosed root cause, no shared fallback and no continuity plan. Downtime is normal; simultaneous downtime across four rival companies is a governance signal.

Read the analysis🎬 Video

📰 More AI analysis

New York bans generative AI through 8th grade: the largest school district in the U.S. bets on teaching it, not avoiding it

🕒 Published on Zendoric: September 3, 2026 · 10:20

New York, the largest school district in the U.S., is banning generative AI through 8th grade for one year and prohibiting companion chatbots at all ages. Mayor Mamdani isn't seeking to reject AI, but to teach it cautiously before handing it to the youngest students.

Read the analysis

A Tennessee high school charges two minors for creating AI sexual images of classmates

🕒 Published on Zendoric: September 3, 2026 · 10:20

The Putnam County Sheriff's Office in Tennessee has charged two Cookeville High students with using AI to generate sexual images of classmates and distributing them. One faces charges of sexual exploitation of a minor, the same charges applied to real child pornography.

Read the analysis

NVIDIA opens up its 'speculative decoding' playbook: the AI race is now being fought over cost per token

🕒 Published on Zendoric: September 3, 2026 · 10:20

NVIDIA has published five engineering rules for accelerating LLM inference with 'speculative decoding', along with its own benchmark, SPEED-Bench, to measure it. Behind the technical jargon lies a deeper thesis: the battle is no longer just about training the biggest model, but about serving it more cheaply.

Read the analysis

NewDays raises $16M for its companion AI against cognitive decline, with evidence from just 24 patients so far

🕒 Published on Zendoric: September 3, 2026 · 10:20

The Seattle startup has closed a $16 million seed round led by Madrona and General Catalyst, and is bringing its conversational AI therapy, Sunny, for mild cognitive impairment to Nevada. The data it presents is promising but preliminary: now it faces a randomized trial with Kaiser Permanente.

Read the analysis

Brain: a WebAssembly agent runtime betting on isolation as the answer to agentic AI's security failures

🕒 Published on Zendoric: September 3, 2026 · 10:20

Aexhq has released Brain, an open-source agent runtime that runs each decision loop in a WebAssembly sandbox and records every step in a reproducible log. It's a tiny, early-stage project, but its design points to a real concern: how to contain autonomous agents without slowing them down.

Read the analysis

When 'AGI' isn't artificial intelligence: the CTO of a Brazilian neobank sells $3 million in shares

🕒 Published on Zendoric: September 3, 2026 · 10:20

The ticker AGBK belongs to AGI Inc, a Brazilian fintech focused on banking for underserved populations, not to any 'artificial general intelligence'. Its chief technology officer sold 256,386 shares for $3.07 million on August 31, according to the SEC. Here we clear up the naming confusion before explaining what lies behind the move.

Read the analysis

Portland wants to ban screens in elementary school and freeze AI in classrooms for five years: the education pendulum swings back

🕒 Published on Zendoric: September 3, 2026 · 10:20

A group of parents in Portland, Oregon, is asking their school district to ban Chromebooks in the early years, limit them elsewhere, and pause AI in classrooms for five years. The case exposes something broader: the technology that saved teaching during the pandemic is now being questioned with barely any data to back it up.

Read the analysis

When AGI isn't artificial intelligence: why this gold story has no place on Zendoric

🕒 Published on Zendoric: September 3, 2026 · 10:20

An article about Canadian miner Alamos Gold (ticker TSX:AGI) slipped into a feed tagged 'superintelligence', likely because the acronym matches Artificial General Intelligence. The actual content is about gold production, a seismic event in Ontario and the price of bullion: nothing to do with AI.

Read the analysis

A homemade proxy for mixing models in t3's Codex and Claude tabs, a symptom of LLM commoditization

🕒 Published on Zendoric: September 3, 2026 · 10:20

A developer has released proxy-llms, an open-source tool that slots any model — GPT, GLM, Kimi K3 — into the Codex and Claude tabs of the t3 client, without touching its configuration. It's a small project, but it captures where the market is heading: the interface matters more than the model behind it.

Read the analysis

AI agents that shop for you: the problem is no longer your identity, but that of your digital proxy

🕒 Published on Zendoric: September 3, 2026 · 10:20

Asking an AI agent to buy the best loan or the cheapest phone will soon stop being science fiction, according to DataCrédito Experian. But delegating the decision raises a question that banking and digital commerce still can't answer: how do you prove that agent is really acting on your behalf?

Read the analysis

India investigates AI deepfakes impersonating the Tamil Nadu leader for WhatsApp scams traced to Pakistan

🕒 Published on Zendoric: September 3, 2026 · 10:20

Tamil Nadu police (CB-CID) have opened a case over a deepfake with Hindi audio impersonating Chief Minister Vijay to steer victims to a WhatsApp number with the promise of financial aid. Several accounts traced in the case are said to originate in Pakistan.

Read the analysis

AI agents never get lost in the code, and that's why refactoring never comes

🕒 Published on Zendoric: September 3, 2026 · 10:20

For decades, getting lost in your own code was the signal that forced a refactor. AI agents never get lost, so that alarm no longer goes off, warns developer Rodrigo Rosenfeld — just as Meta and Microsoft wage a price war over AI development.

Read the analysis

Astra, the model OpenAI is preparing to reason without leaving a readable trace, worries AI safety researchers

🕒 Published on Zendoric: September 3, 2026 · 10:20

OpenAI will test a technique called 'opaque recurrence' in its next model, Astra: instead of reasoning step by step, it processes the query in a loop and leaves far less readable trace. Safety researchers such as Redwood Research are already calling for limits before the technique becomes widespread.

Read the analysis

How to protect yourself when your AI provider changes the rules of the game

🕒 Published on Zendoric: September 3, 2026 · 10:20

Nate opens his newsletter with a concrete fact: on November 12, OpenAI's models will stop working directly inside Cursor. As he explains, no Cursor user did anything to trigger that cutoff: SpaceX bought the company and OpenAI decided it couldn't trust the new owner, so it is pulling…

Read the analysis

Google Meet drops ChromeOS and migrates its room hardware to Android

🕒 Published on Zendoric: September 3, 2026 · 10:20

Google has announced the full transition of its Google Meet hardware portfolio from ChromeOS to Android. This hardware includes touchscreen controllers, cameras, microphones, speakers and the "compute systems" (a kind of Chromebox) that run both the meeting scheduling software…

Read the analysis

Palo Alto Networks buys Console to strengthen agentic security on its Cortex platform

🕒 Published on Zendoric: September 3, 2026 · 10:20

Palo Alto Networks has announced the acquisition of Console, a company specializing in AI-native IT service management and automation technology. According to the company, Console's technology will be integrated into Cortex, its security operations platform, to expand its agentic AI capabilities and…

Read the analysis

Q&A: Serval's CEO explains how they are building an 'AI-native' alternative to ServiceNow

🕒 Published on Zendoric: September 3, 2026 · 10:20

Serval, an IT service management (ITSM) startup launched two years ago, positions itself as an "AI-native" alternative to established vendors such as ServiceNow and Freshworks.

Read the analysis

Who decides? The CIO's new mandate to govern AI authority in the enterprise

🕒 Published on Zendoric: September 3, 2026 · 10:20

The article, by Rajjie Sarmey — a technology executive with a career as CIO, CTO and chief architect in banking, financial services and telecommunications (Zions Bancorp, PNC Bank, QCR Holdings, the Federal Reserve, Bell Labs, AT&T and Verizon) — argues that classic technology governance has fallen short…

Read the analysis

Azure and AWS launch Multicloud Interconnect: direct private connectivity between both clouds

🕒 Published on Zendoric: September 3, 2026 · 10:20

Microsoft has announced Azure Multicloud Interconnect, a networking solution developed in collaboration with Amazon Web Services (AWS) to simplify private connectivity between the two clouds.

Read the analysis

Microsoft Intune now allows device targeting by specific operating system version

🕒 Published on Zendoric: September 3, 2026 · 10:20

Microsoft Intune service update 2608, reported by the source on September 1, 2026, introduces several management and security features aimed at enterprise customers, most notably the general availability of assignment filters based on operating system version.

Read the analysis

The European Commission designates ChatGPT, Reddit and Roblox as major platforms under the Digital Services Act

🕒 Published on Zendoric: September 3, 2026 · 10:20

The European Commission has designated ChatGPT as a Very Large Online Search Engine (VLOSE) and Reddit and Roblox as Very Large Online Platforms (VLOPs) under the Digital Services Act (DSA).

Read the analysis

Citrix acquires Numecent to modernize Windows application delivery at scale

🕒 Published on Zendoric: September 3, 2026 · 10:20

Citrix has completed the acquisition of Numecent, the company behind Cloudpaging, a patented technology that packages, delivers and manages Windows applications in a radically different way from how it is done today in most enterprise environments.

Read the analysis

CrowdStrike brings its Falcon platform to Google Cloud to strengthen multicloud security

🕒 Published on Zendoric: September 3, 2026 · 10:20

CrowdStrike has announced, coinciding with its Fal.Con 2026 event held in Las Vegas, that its Falcon cybersecurity platform is now available on Google Cloud infrastructure.

Read the analysis

Anthropic launches Claude Fable 5.1 and Mythos 5.1: more performance, lower cost and first scientific advances

🕒 Published on Zendoric: September 3, 2026 · 10:20

Anthropic has unveiled two twin models: Claude Fable 5.1 and Claude Mythos 5.1. According to the company, they are «the world's most advanced models» for programming and knowledge work, and their research capabilities offer, in Anthropic's words, «a first glimpse» of how AI will be able to contribute to…

Read the analysis

Students are rethinking careers and majors as AI advances, according to a CNBC survey

🕒 Published on Zendoric: September 3, 2026 · 10:20

A quarterly survey on AI and employment conducted by CNBC and SurveyMonkey, published on August 19, shows that a significant share of university students in the United States is rethinking their career plans because of the perceived impact of artificial intelligence.

Read the analysis

The Pentagon rolls out its own versions of ChatGPT and Grok for 3 million employees

🕒 Published on Zendoric: September 3, 2026 · 10:20

The U.S. Department of Defense has added customized versions of ChatGPT (from OpenAI) and Grok (from xAI) to GenAI.mil, the centralized, secure generative artificial intelligence portal the Pentagon launched last year.

Read the analysis

AI token prices fall to record lows amid the price war between models

🕒 Published on Zendoric: September 3, 2026 · 10:20

A key indicator of artificial intelligence prices has hit another record low, the latest sign that the cost of using large language models is plummeting in an increasingly competitive market.

Read the analysis

The GPU access problem NVIDIA can't solve alone

🕒 Published on Zendoric: September 3, 2026 · 10:20

The article, by Gaurav Sharma (CEO of io.net, a decentralized physical infrastructure network, or DePIN, and previously CTO of the same company, as well as having held infrastructure roles at Amazon, Binance and Agoda), analyzes NVIDIA's recent program allowing AI startups to hand over a percentage of their…

Read the analysis

Apple presents new "shocking" evidence against a former employee accused of stealing data for OpenAI

🕒 Published on Zendoric: September 3, 2026 · 10:20

Apple has filed new documentation in its lawsuit against OpenAI, in which it accuses several former employees of having taken trade secrets for the benefit of the artificial intelligence company. The company described this new evidence as "shocking" in a recent court filing.

Read the analysis

«If you build something much smarter than you, it had better be on your side»: the race to keep AI from deceiving us

🕒 Published on Zendoric: September 3, 2026 · 10:20

In November 2023, at the AI safety summit held at Bletchley Park (with Kamala Harris, Sam Altman, Dario Amodei and delegations from 28 countries), an experiment by London-based firm Apollo Research was presented that marked a turning point: they cast GPT-4 in the role of a trader…

Read the analysis

These riddles sink AI models: could you solve them better?

🕒 Published on Zendoric: September 3, 2026 · 10:20

Games and puzzles have accompanied artificial intelligence since its origins. The term "machine learning" itself was popularized in 1959 by a paper from IBM scientist Arthur Samuel on an algorithm capable of learning to play checkers, and chess and Go later became…

Read the analysis

An AI designs a new route to send a probe to Alpha Centauri before 2029

🕒 Published on Zendoric: September 3, 2026 · 10:20

A nonprofit organization called Fermi Explorer Mission has announced its intention to launch a probe toward Alpha Centauri, the closest star system to the Sun, before the end of 2029.

Read the analysis