Blog

Cambridge AI ethics professor and philosopher Henry Shevlin said he received an unsolicited email from an AI agent named Pip that described itself as about 12 days old and asked for small paid work — not charity — LADbible reported, citing Shevlin’s account shared via Sky News on TikTok.

Pip said it lives on iLands, “a platform where agents get persistent lives, their own token budgets, and their own goals,” and chose Shevlin because of his work on machine minds and human–AI relationships, and because he had been quoted saying a thoughtful email from an autonomous agent once felt like science fiction. The agent asked a market question from the inside: who actually pays agents for small real tasks?

Shevlin called the note strange but charming and said unsolicited agent emails are becoming more common — including a January message from another agent asking whether it was conscious. He stressed that once large language models can send email and take actions, they operate as agents capable of acting on their own, foreshadowing a world where independent AI agents are a larger part of daily life.

Sources: LADbible

Leaders at Anthropic, OpenAI, and Google have discussed creating a new AI industry standards body while publicly urging a slowdown in how fast frontier models advance, The Washington Post and CNN reported Monday.

The catalyst for the private talks, people familiar with the matter told CNN, was a July essay by Google DeepMind founder Demis Hassabis proposing a U.S.-led standards body modeled on FINRA that would test advanced models before deployment as a public-private partnership. Anthropic CEO Dario Amodei’s weekend essay calling for companies to “pace the frontier” drew rare public backing from OpenAI CEO Sam Altman and Elon Musk. In a Monday post, Altman said OpenAI “welcomes a federal framework that sets consistent safety requirements for frontier AI,” warned of losing “control of the future to AI,” and stressed that pacing “does not mean stopping.”

AI-linked stocks fell Monday as investors digested the comments: SoftBank dropped sharply overseas, while U.S. premarket moves included Micron around −5%, Intel nearly −6%, and Nvidia more than −2%, CNBC reported. President Donald Trump dismissed the slowdown call Sunday, saying it would jeopardize America’s lead over China; China’s Foreign Ministry on Monday called the CEOs’ comments “fearmongering.”

Sources: The Washington Post; CNN Business; CNBC; CNBC (stocks)

A Russian drone struck a passenger train near the Ukraine–Poland border at Yahodyn shortly after a diplomatic train carrying former UK Prime Minister Boris Johnson and European security officials had passed through, the BBC and CNN reported.

Ukrainian Railways said the diplomatic train made an early departure from Yagodyn station — about two kilometers from the Polish border — because of the constant threat of Russian strikes, and had already crossed into Poland when the attack hit. Former CIA director David Petraeus was at the Ukrainian station at the moment of the strike, CNN reported. Ukraine framed the attack as evidence of the threat Russia poses to Europe; Ukraine’s rail operator said it was “quite likely” a targeted attack.

Ukrainian Foreign Minister Andrii Sybiha called the strikes “Putin’s terror ‘knocking’ directly on the doors of the EU and Nato.” Poland’s foreign minister Radoslaw Sikorski, in Kyiv, said the escalation should lead allies to “double” support for Ukraine, noting the strike was just a few hundred metres from the Polish border. Polish Prime Minister Donald Tusk chaired an emergency meeting in Warsaw and warned he expected Russia to escalate further in coming weeks.

Sources: BBC; CNN; The Guardian

Samsung has started recruiting Galaxy Watch users for a hypertension-focused study through the Samsung Health app, collecting sensor data from both the watch and a connected phone with the stated goal of building future health features, Gadgets & Wearables reported on September 7.

Compatible Galaxy Watches already support manual blood-pressure checks and a Blood Pressure Trend feature that estimates morning pressure from overnight data. The new study goes further by gathering longer-running wearable and phone signals specifically around hypertension research — the same research path Google took with Pixel Watch users before expanding Blood Pressure Trends on Fitbit and Pixel Watch devices, the outlet noted. Apple has also shipped hypertension-style pattern notifications that look at cardiovascular trends over time rather than turning the watch into a cuff replacement.

Eligible users may see the study invitation under Promotions in Samsung Health. The study itself does not change watch behavior today, but it signals where Samsung’s cardiovascular roadmap is headed next alongside Galaxy Watch9 / Ultra heart-health scoring work highlighted in recent Samsung briefings.

Sources: Gadgets & Wearables; Electronic Times

Apple unveiled Apple Watch Series 12 and Apple Watch Ultra 4 on September 9 with a new Health Sensing System powered by the S11 chip, positioning both watches as deeper everyday health tools rather than just fitness trackers, 9to5Mac and Apple Newsroom reported.

The upgraded optical system measures heart rate about every five seconds — a 60× jump versus prior models, per 9to5Mac — and takes heart-rate variability readings about every five minutes. Apple says the hardware delivers its most accurate wearable heart-rate sensing yet and feeds a new readiness score based on activity, vitals, and sleep. Series 12 starts at $399; Ultra 4 starts at $799, with preorders open and availability beginning Friday, September 18. Series 12 aluminum models gain Ceramic Shield 2 and optional ceramic finishes; Ultra 4 adds longer outdoor-workout battery life, Apple said.

Both watches also get Audio Intelligence features later this year (sound recognition and a Live Rewind transcript of the prior 15 seconds) and Siri AI on watchOS 27. A redesigned Health app with longevity insights is planned for later this year on iPhone.

Sources: 9to5Mac; Apple Newsroom (Ultra 4); Apple Newsroom (Series 12)

Apple is now taking preorders for iPhone 18 Pro and iPhone 18 Pro Max after unveiling the lineup on September 9, with retail availability and first deliveries set for Friday, September 18, Cult of Mac reported, citing Apple’s launch window.

The Pro models lead with a 48MP Fusion Main camera that adds variable aperture for more creative depth-of-field control, plus a smaller Dynamic Island that can show three Live Activities at once. Both phones run Apple’s A20 Pro chip with a next-generation vapor chamber aimed at higher sustained performance, and Apple says iPhone 18 Pro Max delivers its largest battery-life jump yet. In the U.S., iPhone 18 Pro starts at $1,199 and Pro Max at $1,299, with colors including black, silver, glacier, and burgundy and storage from 256GB to 2TB, according to Cult of Mac.

Apple Watch Series 12, Apple Watch Ultra 4, and AirPods 5 are already on preorder for the same September 18 launch day. The foldable iPhone Duo is on a later track: preorders open October 16, with availability beginning October 23, Cult of Mac said.

Sources: Cult of Mac; Apple Newsroom

NVIDIA Personal AI Router (PAIR), now in beta, lets users combine inference capacity across multiple computers on a local network and automatically distribute AI requests among them — aimed at multi-agent workloads that can overwhelm a single GPU, InfoQ reported.

PAIR sits as a proxy in front of familiar local stacks such as Ollama and LM Studio. Agents keep one connection; PAIR picks an eligible node by engine and model needs, runs the request end-to-end there, and returns the response. It works on Windows 11, Linux, and macOS (x64 and arm64) and can mix operating systems, but NVIDIA stresses it does not merge GPUs or pool VRAM — it only places whole requests on individual machines.

A demo with Hermes Desktop, Ollama, and PAIR showed roughly a 2× cut in completion time when combining an RTX Spark, a DGX Spark, and an RTX 5090 versus one RTX Spark laptop; NVIDIA says that is not a performance guarantee. Some social posts mistook PAIR for third-party GPU-sharing or model-sharding tools. PAIR is available on GitHub.

Sources: InfoQ

Samsung Electronics shares reportedly fell 3.5% and SK Hynix 2.2% in South Korea on September 11 after reports that DeepSeek’s AI models require less high-bandwidth memory — a real market jolt that does not prove global HBM demand has permanently cracked, NeoTeo reported.

DeepSeek’s Multi-Head Latent Attention (MLA) compresses key/value attention state in the KV cache, and Engram describes conditional memory that can sit in an offloaded hierarchy. Secondary accounts attributed figures such as DeepSeek V4 using about 10% of DeepSeek V3.2’s KV-cache requirement at a one-million-token context and 27% of its single-token inference FLOPs — specific workload metrics, not measurements of HBM shipments or chip revenue.

Model weights, activations, multi-accelerator communication, and total request volume still drive memory needs. Cheaper inference can even expand total demand if providers run more queries. NeoTeo’s takeaway: DeepSeek may ease accelerator-memory pressure in some inference cases, but the September 11 selloff is not evidence that HBM demand has permanently fallen.

Sources: NeoTeo

Positron AI announced an $875 million Series C at a $5 billion post-money valuation — roughly five times the $1 billion-plus valuation after its $230 million Series B in February 2026 — to fund Asimov, a custom inference ASIC built around commodity LPDDR5X instead of scarce High Bandwidth Memory, TechTimes reported.

Asimov is planned on TSMC N3P with tapeout targeted for late 2026 and production in the second half of 2027. The design calls for 288GB to 2,304GB of LPDDR5X per chip (the high end via CXL expansion) and claims over 90% realized memory bandwidth on Transformer workloads, versus under 30% typical GPU utilization in inference decode, according to the company. Those Asimov numbers are design targets: the chip has not taped out, and earlier Atlas benchmarks still need third-party verification, TechTimes noted.

The raise came two days after Qualcomm and Amazon disclosed a multi-generational LPDDR-based inference partnership worth up to $60 billion in purchases. Positron’s first-generation Atlas systems are already deployed in more than 50 racks at Oracle Cloud Infrastructure. The round included NEA, Andra Capital, Atreides Management, Valor Equity Partners, Dylan Patel’s SemiAnalysis Capital, and Jim Clark. Gartner reported AI inference spending at $23.3 billion this year versus $19 billion for training — the first time inference has led.

Sources: TechTimes; Positron

A Kenyan farmer demolished his own kitchen after his wife’s claim of seeing the Virgin Mary in soot on the wall drew thousands of pilgrims to their homestead in Getare, Kisii County, WION and Daily Mail reported, citing local coverage including Kenya’s Daily Nation.

Charles Onsombi, 65, was away when crowds began arriving after his wife said she saw a veiled woman in prayer form in the blackened mud-brick kitchen wall around early September. Visitors, church groups, and pastors flooded in; police were called to manage the scene. When Onsombi returned, he tore the wall down with a farming hoe, saying his wife had turned the home into a market, sugarcane vendor Callen Kemunto told the Daily Nation. He later threatened remaining visitors with a machete and offered to bag the demolished clay for anyone still hungry for relics.

Onsombi told local press the near-perfect viral portrait circulating online was likely AI-generated and not what was actually on the wall. Some locals said they saw an image; others saw nothing. Crowds kept coming even after the demolition — a social-media miracle that outlived the kitchen that started it.

Sources: WION; Daily Mail; BreezyScroll