Zendoric

Zendoric

AI analysis and perspective, every day.

Updated: September 7, 2026
🎙️ Listen to today's podcast

🔥 In focus today

Altman tells the G20 no child will ever outsmart AI: prophecy and lobbying in the same sentence

🕒 Published on Zendoric: September 7, 2026 · 02:54

At the G20 innovation ministerial meeting in North Carolina, Sam Altman claimed a child growing up today will never be smarter than AI and compared rejecting it to having rejected electricity. It is a message aimed at governments, not a scientific finding: prophecy is worth separating from evidence.

Salesforce rebounds on AI: Agentforce starts booking revenue and Anthropic's IPO boosts the value of its bet

🕒 Published on Zendoric: September 7, 2026 · 02:54

Salesforce stock is up more than 80% from its year lows, to $260, after results that belie — for now — the feared "SaaSpocalypse". The other engine: its stake in Anthropic, which is reportedly preparing an IPO valued at up to $2 trillion.

Altman tells the G20 no child alive will outthink AI — but his electricity analogy is the real argument

🕒 Published on Zendoric: September 7, 2026 · 02:54

At the G20 Innovation Ministerial in North Carolina, OpenAI's CEO told ministers that "a child growing up today will never be smarter than AI" and compared refusing the technology to refusing electricity a century ago. The first claim is unfalsifiable rhetoric aimed at policymakers; the second is a genuinely useful frame — provided you remember how long electrification actually took.

Huang says 'AGI has arrived' — the man selling the GPUs shouldn't get to define the finish line

🕒 Published on Zendoric: September 7, 2026 · 02:54

Nvidia's CEO congratulated OpenAI on its new Astra model and declared that artificial general intelligence is here, noting it was trained on his chips. Critics say he offered no definition and no evidence — and OpenAI's own CEO has called AGI "an irrelevant marketing term." The disagreement is not semantic; it decides how the technology gets governed.

A humanoid robot ran 100m faster than Bolt's record — an actuator milestone, not an athletic one

🕒 Published on Zendoric: September 7, 2026 · 02:54

At the Second World Humanoid Robot Games in Beijing, Tiangong Ultra covered 100 metres in 8.86 seconds, ahead of Usain Bolt's 9.58s world record, while other machines cleared 2.88m in the vertical jump. The numbers are real and the comparison is misleading — and both facts matter.

OpenAI lifts free-tier text limits: distribution, not capability, is the battleground now

🕒 Published on Zendoric: September 7, 2026 · 02:54

OpenAI has removed message caps and the quota system for text chats on ChatGPT's free tier, so anyone can keep chatting without a subscription. It reads less like generosity and more like a land grab: when models converge, the moat moves to who owns the default habit.

AGI 'has arrived' on OpenAI's own harness — the benchmark's authors measured 62.7%

🕒 Published on Zendoric: September 7, 2026 · 02:54

OpenAI reported 99.9% on ARC-AGI-3 for Astra; the organization that built the test published 62.7% from its own reference harness. Both numbers are real, which is what makes the 37-point gap a governance problem rather than a scandal. And as of publication, the two most-watched independent indexes still had no entry for the model.

OpenAI reportedly says its AI agents now outwork its humans 3-to-1 — the metric is the real story

🕒 Published on Zendoric: September 7, 2026 · 02:54

According to the reported internal figures, work performed by AI agents inside OpenAI already runs at roughly three times the output of its human staff. It is a striking number — and an unfalsifiable one until someone defines what "output" means. Our read: treat it as a signal about how the company measures itself, not as proof of a labor threshold crossed.

📰 More AI analysis

Sylvester secures BCRF funding to bring an AI 'companion' to breast cancer survivor follow-up

🕒 Published on Zendoric: September 7, 2026 · 02:54

Miami's Sylvester Comprehensive Cancer Center has received a Breast Cancer Research Foundation award to build My Wellness Support, an AI assistant that monitors breast cancer survivors' symptoms between appointments. It is a development grant, not an already validated product: that's the nuance.

Jensen Huang says AGI is already here: the man selling the machines also sells the story

🕒 Published on Zendoric: September 7, 2026 · 02:54

Nvidia's founder and CEO claims artificial general intelligence is already a reality, according to Yahoo Finance and BeInCrypto. The detail worth keeping in view: his company sells the chips that make training and running those very systems possible.

MaskShift: a local, NPM-free coding agent that is 'permissive by default', barely noticed on Hacker News

🕒 Published on Zendoric: September 7, 2026 · 02:54

A developer has posted MaskShift on Hacker News, a local coding agent with no NPM dependencies that gives tools even to models without a function-calling API. With one point and zero comments, the project sums up the craze for the permissive-by-default agent that 'does everything'.

Bobbin: a 100% local AI pentesting agent that also publishes its failures

🕒 Published on Zendoric: September 7, 2026 · 02:54

Bobbin is a coding agent for small models running locally via Ollama: the code never leaves the machine, so it can audit NDA-bound projects no cloud LLM could touch. Its author publishes thousands of failed runs alongside the successful ones, a benchmark transparency rarely seen in the sector.

Bill Gates: AI will be history's greatest equalizer or its worst injustice, and there is no plan to decide which

🕒 Published on Zendoric: September 7, 2026 · 02:54

Bill Gates warns on his blog that AI could be the greatest equalizer of opportunity ever created or the worst source of injustice, and attributes the risk to the absence of a plan to manage the transition amid geopolitical and economic pressure. The Microsoft founder rules out that such a roadmap will arrive in time.

The distribution of Anthropic's $1.5 billion runs into publishers and agents claiming more than their due

🕒 Published on Zendoric: September 7, 2026 · 02:54

Authors are complaining on social media that publishers such as HarperCollins and several literary agencies are claiming payments from Anthropic's landmark settlement over pirated books, even in cases where they no longer hold any rights to those works. The first major collective payout for training AI on stolen content reveals how unequal it is, in practice, to defend an individual right against an industry with more lawyers.

New York bans generative AI in elementary schools for a year, and the debate already divides those who ran its own schools

🕒 Published on Zendoric: September 7, 2026 · 02:54

New York is banning generative artificial intelligence in elementary school classrooms for at least this school year, as announced by Mayor Zohran Mamdani. Former schools chancellor David Banks calls it "putting your head in the sand": children already use AI, he says, and banning it in class does not remove it from their lives.

When your neurodivergent child's therapy robot 'dies' because the startup runs out of cash

🕒 Published on Zendoric: September 7, 2026 · 02:54

Moxie, a 15-centimeter companion robot designed to teach social skills to neurodivergent children, stopped working when its maker, Embodied, ran out of funding and shut down its servers. Thousands of children who treated it as a friend faced a sudden grief inadvertently designed by a business model.

meclaw: a Rust binary builds agents with a kernel sandbox, and its creator deliberately switches off self-improvement

🕒 Published on Zendoric: September 7, 2026 · 02:54

An independent developer has released meclaw, a system that builds AI agents as folders isolated by the Linux kernel, without the 'call the model and repeat' loop that dominates agentic frameworks. The most revealing part: it includes the pieces for recursive self-improvement and deliberately leaves them disconnected.

Kraken lets you bet on OpenAI and Anthropic with leverage before there is any stock to buy

🕒 Published on Zendoric: September 7, 2026 · 02:54

Kraken has launched leveraged perpetual futures (up to 5x) on the private valuations of OpenAI and Anthropic, with no expiry and settled in dollars. They are not shares: they carry no voting rights or dividends, and their reference price is generated by the derivatives market itself, not by a real public listing.

JPMorgan and Vista Equity: the agentic AI business is not in the models, it's in the software that distributes them

🕒 Published on Zendoric: September 7, 2026 · 02:54

A JPMorgan Private Bank webcast with Vista Equity Partners reveals where private capital is placing its AI investment thesis: not on the large models, but on the enterprise software that turns them into agents working 24/7. The underlying figure: more than 95% of software companies remain private.

Michigan State creates an AI minor and makes AI literacy mandatory university-wide, not just in engineering

🕒 Published on Zendoric: September 7, 2026 · 02:54

Michigan State University is launching a 22-credit minor in artificial intelligence this year along with two new AI literacy courses open to any student, funded by a $5 million fund. The telling detail: one of those courses failed twice for lack of enrollment in 2025 and is now full.

GPT-6 Astra and the work you no longer see being done

🕒 Published on Zendoric: September 7, 2026 · 02:54

The briefing's starting point is a simple idea: throughout the history of computing, the machine waited for the human to decide what to do and when. It performed that specific task and stopped.