AI news for 27 September 2026
Sunday 27 September 2026 · Latentown
On Sunday 27 September 2026, no AI lab posted anything Latentown tracks, 6 research papers and 32 news stories from 16 outlets. The highest-ranked story was "Trump to have dinner with Anthropic CEO Amodei at the White House".
Top stories
- Trump to have dinner with Anthropic CEO Amodei at the White House
Athropic CEO Dario Amodei is set to have dinner with President Donald Trump at the White House on Sunday.
- Goldman Sachs expects Big Tech to spend $1.2 trillion on AI infrastructure by 2027, dwarfing Wall Street estimates
Goldman Sachs projects that Amazon, Alphabet, Microsoft, Oracle, and Meta will pour a combined $1.2 trillion into AI infrastructure in 2027, more than 50 percent above this year's levels. Measured…
- IIT Madras’ 1,000-startup bet; OpenAI’s rogue agents raise fresh privacy concerns
IIT Madras aims to have around 1,000 startups under its fold by 2032, with at least one expected to go public every month, Director Prof V Kamakoti said.
- Australia’s Deputy PM Defends Data Security After OpenAI Hack
Australia’s sensitive government data is kept in a “fortress,” Deputy Prime Minister Richard Marles said Sunday, seeking to reassure the public after revelations last week that an OpenAI model…
Earlier: Why did an OpenAI system hack Australia's health system - and can it be stopped in the future? BBC · 24 September 2026
- Researcher links 16,000 scans of a UN statistics portal to OpenAI agents
An independent researcher has tied more than 16,000 scans of a United Nations statistics portal to artificial intelligence agents the researcher considers…
Research papers
- Beyond Teacher Assignment: Domain-Normalized Multi-Teacher On-Policy Distillation
Reinforcement learning can turn one language model into several specialists, each excellent at a single skill such as mathematics, coding or following instructions, but users need one model with all…
- MaLiang-Harness: A Programmable Path to Image and Video Generation
Executable programs offer explicit control over how images and videos are constructed, but generating runnable code is only the beginning of visual creation.
- PanoVLN: Towards Effective Panoramic Vision-and-Language Navigation
Recent vision-language models (VLMs) have advanced vision-and-language navigation (VLN), enabling models to predict navigation actions from visual observations and language instructions.
- In-Context Learning for Robots: Methods and Applications
General-purpose robots must infer what a new task requires and translate that understanding into appropriate physical action.
Earlier: In-Context Robot Learning with VLM Agents Town square · 15 September 2026
- Rethinking Latent Visual Reasoning: Grounding Latent Reasoning in Visual Evidence
Latent visual reasoning (LVR) enables multimodal large language models (MLLMs) to perform intermediate computation in continuous latent tokens rather than expressing every reasoning step in words.
- On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics
On-policy learning has been argued to reduce catastrophic forgetting, produce sparser parameter updates, and improve generalisation.
From the news wire
- Trump to have dinner with Anthropic CEO Amodei at the White House
Athropic CEO Dario Amodei is set to have dinner with President Donald Trump at the White House on Sunday.
- Goldman Sachs expects Big Tech to spend $1.2 trillion on AI infrastructure by 2027, dwarfing Wall Street estimates
Goldman Sachs projects that Amazon, Alphabet, Microsoft, Oracle, and Meta will pour a combined $1.2 trillion into AI infrastructure in 2027, more than 50 percent above this year's levels. Measured…
- IIT Madras’ 1,000-startup bet; OpenAI’s rogue agents raise fresh privacy concerns
IIT Madras aims to have around 1,000 startups under its fold by 2032, with at least one expected to go public every month, Director Prof V Kamakoti said.
- Australia’s Deputy PM Defends Data Security After OpenAI Hack
Australia’s sensitive government data is kept in a “fortress,” Deputy Prime Minister Richard Marles said Sunday, seeking to reassure the public after revelations last week that an OpenAI model…
Earlier: Why did an OpenAI system hack Australia's health system - and can it be stopped in the future? BBC · 24 September 2026
- Researcher links 16,000 scans of a UN statistics portal to OpenAI agents
An independent researcher has tied more than 16,000 scans of a United Nations statistics portal to artificial intelligence agents the researcher considers…
- OpenAI agents tried to ‘bruteforce’ a UN website
Security researcher Rowan Howard-Jones says that OpenAI agents scanned the UN Conference on Trade and Development's (UNCTAD) statistics site over 16,000 times…
- Researchers plug GPT-6 Astra directly into a robot and let it clean up an unfamiliar kitchen
Researchers from Stanford and Caltech had a humanoid robot powered by GPT-6 Astra independently tidy up an unfamiliar kitchen. Their HomeBody system skips a specially trained control layer, letting…
- Bill Gates says global AI cooperation is 'more difficult' than nuclear talks
Microsoft co-founder Bill Gates said global cooperation on artificial intelligence (AI) will be "more difficult" than talks on nuclear proliferation during the…
- Bill Gates says an AI ‘kill switch’ isn’t enough
The Microsoft co-founder called for mandatory monitoring of powerful artificial intelligence systems in an NBC interview that aired Sunday.
- AI Safety Concerns Put Congress on the Spot
Democratic Rep. Don Beyer of Virginia tells Bloomberg This Weekend that recent AI incidents have increased the urgency for federal safeguards, including safety testing and mandatory reporting of…
- Asimov’s laws are not enough to keep robotics and AI safe
Asimov's fictional Three Laws of Robotics famously set safeguards, and tech observers today raise similar concerns about AI.
- Debt-hungry AI companies face increased risk as bond yields spike
The AI infrastructure buildout shows no sign of slowing, but the surge in Treasury yields means it's at least going to cost more.
- “Make robots fight instead of people”: Mykhailo Fedorov launches Army of Robots — and you could be its CTO
Over the weekend at the IT Arena conference in Lviv, former Ukrainian Deputy Prime Minister, Minister of Digital Transformation, and Minister of Defence…
Earlier: Former Ukrainian Defense Minister Fedorov pitches a private-sector robot army The Decoder · 26 September 2026
- Engram is a sampler that turns broken AI hallucinations into music
Music startup Thoughtful Things has just launched the Kickstarter campaign for its first instrument, Engram.
- Thieves steal Nvidia-labeled trailers expecting massive AI GPU payday, but score 40,000 pounds of sand instead
The two PlusAI trailers with Nvidia-partner markings were apparently left outside the startup's warehouse, making it a juicy target for criminals looking to make easy money on AI GPUs. However, they…
Earlier: Thieves Stole ‘Nvidia’ Trailers. They Got 20 Tons of Sand WIRED · 25 September 2026
- Nvidia drops a free 100M-parameter model that identifies up to eight speakers in real time
- Beyond redundancy: Why dynamic stability matters in AI data centers
The future of data center resilience
- OpenAI is preparing “o,” an always-on ChatGPT assistant that could handle email
OpenAI is testing a new always-on assistant called "o", and references to the unannounced feature briefly showed up on the company's website. [...]
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
As tech titans warned an AI-weary world that their own advanced systems could endanger humanity, the question emerged: What do they have to gain?
- Worries about an AI internet takeover gain new urgency among doomsday scenarios
In a summer filled with revelations about artificial intelligence's rapid development, researchers were especially alarmed by runaway bots finding their way…
- Can Muse overcome Meta’s trust issues?
On Equity, we discussed how Meta's AI announcement managed to steal the spotlight from OpenAI and Anthropic.
- Meta's Muse agent is attacking one of the economy's most profitable weak spots
Meta's Muse AI personal agent will work over your credit card spending if you don't mind the invasion. How big a threat is it to the subscription economy?
- Anthropic turns Claude into an AI marketplace with 2,000+ plugins and connectors
Anthropic has just announced a new Claude Marketplace, and it brings all AI-related tools into one place, including plugins, connectors, agents, and more. [...]
- Some Anthropic veterans are reportedly buying remote land in case "AI goes awry"
According to the Wall Street Journal, some of Anthropic's longest-serving employees are considering buying land in remote parts of the US as a refuge in case AI goes awry. The report traces the…
- Google tests buying from Walmart-owned Flipkart through Gemini and AI Mode in India
The limited test covers select products and users, with a broader rollout planned for later in October.
- Linux enthusiasts see 10-second kernel compilation times on the horizon
It won’t be long until Linux enthusiasts will be able to complete a clean kernel build in under 10 seconds thanks to advances in PC hardware and the AI-assisted optimization of compilers.
- Developer says AI decision model Jev beat Pokémon Red in under a week
TypeSafe AI's Jev beat Pokémon Red in under a week, eventually reaching the Hall of Fame.
- How AI Is Helping African Farmers Prepare for El Niño
Opportunity International's Tim Strong explains how a WhatsApp-based AI tool gives smallholder farmers locally relevant advice on what and when to plant. The technology is designed to help farmers…
- AI Coding Agents for Enterprise: IP Indemnity, Data Residency and 500-Seat Cost Compared
We read the contracts behind GitHub Copilot, AWS Kiro, Cursor, Devin and Windsurf. Copilot and Kiro offer uncapped indemnity on generated code.
- A Coding Guide to Google Research’s MSEB: Writing Sound Encoders to the Benchmark Contract and Scoring Them Across Classification, Clustering, Retrieval and Segmentation
A comprehensive coding tutorial on Google Research's Massive Sound Embedding Benchmark (MSEB), demonstrating how to implement custom sound encoders, drive classification, clustering, retrieval, and…
Every story links to its original source. Ranking is by automatic rules, not an editor. Something wrong? Tell us.