AI news for 22 October 2026
Thursday 22 October 2026 · Latentown
On Thursday 22 October 2026, Latentown tracked 5 posts from 2 AI labs, 5 research papers and 45 news stories from 23 outlets. OpenAI was the busiest lab with 4 posts. The highest-ranked story was "Build more responsive apps with the Decisions API".
Top stories
- Build more responsive apps with the Decisions API
Fast, focused, low-cost decisions for any application.
- Anthropic Takes On OpenAI, Meta With New Proactive AI Agent Claude Duet
Anthropic has unveiled Claude Duet, a proactive AI agent designed to monitor connected apps, prioritise tasks, and carry out work…
- Anyverse Dynamics Open-Sources MWA-Phoenix, a 7B On-Device World Action Model for Robots
Anyverse Dynamics released and open-sourced MWA-Phoenix, a 7B latent world action model that uses only RGB input and reports leading scores on six robot…
- Huawei Gets $2.3 Billion Offshore Loan as Chip Spending Soars
Huawei Technologies Co. has secured a 15 billion yuan ($2.25 billion) loan from Chinese banks, according to a person familiar with the matter, as spending on…
- OpenAI to make its most advanced AI models available in Google Cloud
OpenAI Group PBC today announced plans to make its latest large language models accessible in Google Cloud.
From the AI labs
OpenAI
- Build more responsive apps with the Decisions API
Fast, focused, low-cost decisions for any application.
- Uber brings ChatGPT Work to all employees with OpenAI
Uber is rolling out ChatGPT Work to more than 30,000 employees, helping teams research, analyze data, create presentations, and complete complex workflows.
- What FDE means at OpenAI
What forward deployed engineering means at OpenAI
- OpenAI Python SDK 3.36.1 released
A new release of the OpenAI Python SDK. Changes include: api: honor per-call WebSocket query settings.
Google DeepMind
- Beyond tree cover: mapping the world’s forest types
A global map categorizing every 10-meter patch of land on Earth into distinct forest types.
Research papers
- RynnUnify: Towards Unified Embodied Brain
We introduce RynnUnify, a 16B unified embodied model that combines strong embodied understanding with long-horizon visual planning through interleaved video--language reasoning.
Earlier: SenseNova-U1.5: Towards Native Unified Visual Intelligence Town square · 11 September 2026
- Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution
LLM agents are increasingly used to solve real-world problems, and their effectiveness depends on both the underlying models and the harnesses that support their operation.
- The science behind TorchNeuron, Amazon’s new PyTorch backend for AWS Trainium chips
Runtime optimizations and a thorough evaluation process driven by agentic AI help ensure that models written in PyTorch will execute efficiently and reliably on Trainium systems.
- Flint-interactive: A language for creating interactive data visualizations
Flint-interactive gives AI agents a language for creating interactive visualizations. Click a chart, filter data, highlight related content, and explore beyond the view the author originally chose.
- Making MLLM Latents Matter: Factor-Aware Predictive Learning for Cinematic Video Generation
Cinematic video generation requires both temporal inference from limited inputs (i.e., sparse rather than frame-by-frame guidance) and factor-aware control of character, scene, action, and camera.
From the news wire
- Anthropic Takes On OpenAI, Meta With New Proactive AI Agent Claude Duet
Anthropic has unveiled Claude Duet, a proactive AI agent designed to monitor connected apps, prioritise tasks, and carry out work…
- Anyverse Dynamics Open-Sources MWA-Phoenix, a 7B On-Device World Action Model for Robots
Anyverse Dynamics released and open-sourced MWA-Phoenix, a 7B latent world action model that uses only RGB input and reports leading scores on six robot…
- Huawei Gets $2.3 Billion Offshore Loan as Chip Spending Soars
Huawei Technologies Co. has secured a 15 billion yuan ($2.25 billion) loan from Chinese banks, according to a person familiar with the matter, as spending on…
- OpenAI to make its most advanced AI models available in Google Cloud
OpenAI Group PBC today announced plans to make its latest large language models accessible in Google Cloud.
- Dream Labs raises $90M and launches Dream API
Dream Labs says its Dream-1 AI model works across robots from different manufacturers and will allow robots to transfer their learning.
- OpenAI says ChatGPT Sol helped uncover critical PHP flaws that could affect 70% of websites
OpenAI says researchers using GPT-5.6 Sol uncovered two critical PHP vulnerabilities with severity scores of 9.2 and demonstrated working exploits, after which…
- CureBay raises Rs 100 Cr in Pre-Series C round led by IvyCap Ventures
CureBay said it will use the funds to expand its operations across India, strengthen its healthcare delivery network, and develop its technology and AI…
- Former OpenAI head of data centers joins Nvidia to help lead DSX Platform team
Chris Malone announced new role two months after OpenAI departure was reported
- Körber takes stake in bootstrapped German robotics startup NAiSE
Industrialtech company Körber is acquiring a stake in NAiSE, a Germany-based software company specialising in vendor-agnostic fleet management.
- Nvidia's Jensen Huang says he sees no sign of a 2029 memory glut: 'You can always use more'
Nvidia's Jensen Huang told CNBC that a surplus of memory supply is unlikely by 2029, as memory stocks have taken a beating in the past week.
- 1Password lets AI agents use your passwords without seeing them
1Password is introducing a new feature that allows AI agents to sign into your accounts without actually seeing or handling your passwords.
- Is training AI on song lyrics fair use? US court weighs question as India proposes mandatory licensing
Anthropic has asked a US court to rule that training Claude on copyrighted song lyrics is fair use, while music publishers argue that AI-generated lyrics…
- '500 new vulnerabilities' found in government, says AI minister
- Government tells TV channels to limit AI to ‘assistive capacity’. Does that rule out AI anchors?
From Aaj Tak’s Sana to India Today’s Sutra, AI anchors are spreading across Indian television.
- Hackers Exploited the PHP Vulnerabilities Days Before They Were Patched
Amazon Web Services logged attempts to leverage the internet-melting PHP2Shell bugs four days ago.
- Southeast Asia’s sellers spent a decade learning to sell to people. Now they have to sell to AI
In February 2026, Sea Limited struck a partnership with Google on “agentic shopping prototypes.” By June, Shopee had become the first Southeast Asian…
- Barcelona's Masia closes €18M debut fund and opens new home for frontier tech founders
Masia has closed its €18M debut fund to invest in early-stage AI and frontier tech companies with a Barcelona connection, and opened Masia House to bring its…
- Ai2 releases OlmoEarth v1.3 state-of-the-art AI-ready satellite embeddings for the whole planet
Ai2, the Allen Institute for Artificial Intelligence, today released a satellite map embedding model called OlmoEarth v1.3 and precomputed annual embeddings…
- Binny Bansal-Backed ShopOS Launches AI Creative Platform Sloosh For Individuals
Binny Bansal-backed AI startup ShopOS has launched Sloosh, a self-service platform that makes the image and video generation technology it…
- Tsinghua AIR Spinout Tairex Brings Multi-Specialty AI Doctor Avatars Into Outpatient Clinics
Tairex, incubated by Tsinghua AIR, launched MHAC, a framework that lets a clinic doctor call in AI avatars of other specialists for advice, while human doctors…
- BAAI's FlagOS 2.2 Expands Its Cross-Chip Operator Libraries to 1,454 Operators
FlagOS 2.2 from the Beijing Academy of Artificial Intelligence grows its nine operator libraries to 1,454 operators and reports about 27% faster operators…
- IBM AI Releases Granite Speech 5.0 2B for Multilingual Transcription and Translation
IBM released Granite Speech 5.0 2B, an Apache 2.0 speech recognition and translation model.
Earlier: Granite Speech 5.0: Transcribe, Translate, and Attribute more Accurately and Faster Hugging Face · 21 October 2026
- Nokia Profit Beats Estimates as Data Centers Fuel Growth
Nokia Oyj’s third-quarter profit beat analyst expectations, driven by rising demand for data center networking equipment as investment in artificial…
Earlier: Samsung forecasts record third-quarter profit of $80 billion as AI boom fuels chip demand CNBC · 8 October 2026
- Perplexity and NVIDIA Researchers Introduce JOLT: a Joint Teacher-Student Reinforcement Learning Method
Perplexity and NVIDIA introduced JOLT, a post-training method that trains one model as both teacher and student.
- China Telecom's TeleAI Spins Out Gamma Robotics to Build Self-Evolving Humanoid Robots
Gamma Robotics, the first market-oriented company incubated by China Telecom's TeleAI, has launched in Shanghai to develop self-evolving humanoid robots and…
- China’s Unitree unveils ‘guardian’ robot that can disarm attackers, lift cars
A knife-wielding intruder is confronted.
- Recursive shows first result of its AI-driven research process with Flint
The AI startup Recursive is unveiling Flint, its first language model, a 40-billion-parameter MoE for local coding developed by former DeepMind researchers.
- Meta Q3FY26 earnings call: From shopping to smart glasses, Meta expands its Muse AI push
Meta says over 10 million people have set up Muse as it expands AI-powered shopping, advertising, business tools and smart glasses integration, while facing…
- Yann LeCun and Arthur Mensch back Zurich robotics startup in $30m pre-seed
- Spotify’s current AI bet is about being the music service that agents choose
Spotify co-CEO Gustav Söderström explains why the rise of AI agents could be good news for the music service.
- Unitree developing AI system to let humanoid robots improve themselves
Unitree Robotics is developing a “self-evolving” feedback loop designed to allow humanoid robots to write code, run simulations and improve their real-world…
- Why Engineers Should Brave Capitol Hill
When San Antonio–based software engineer Nikhil Jathar first went to Washington, D.C., to advocate for federal support for technology education and research…
- Meet Mori, a Cute Robot Mushroom That Doesn’t Really Say (or Do) Anything
Step one in convincing people to live with robots with cameras on their faces?
- Making the case for context-aware AI
Enterprise AI can generate recommendations, automate tasks, and help organizations make decisions faster.
- Ensemble forecasting with IBM Granite time-series models
- Elbilassy: Tech Selloff Healthy After Nasdaq Record High
Stocks fell as megacap earnings failed to reassure investors about the rebound in Artificial Intelligence trade.
- Barclays to Cut 5,000 Jobs as AI Starts to Create Efficiency
Barclays Plc said it’s cutting 5,000 jobs as part of an effort to reduce costs, adding that artificial intelligence is starting to create efficiency.
- Who referees the AI agents? Enterprises aren’t sure, new report shows
Ask a room of enterprise executives whether they can keep their AI agents in line and most will nod.
- Ireland’s AI scene has buzz. Now it needs breakouts
- Will AI agents drive customers away from travel platforms and its impact on ixigo
ixigo plans to build its own AI travel agent while partnering with third-party agents to handle bookings.
- How I decide what to fix in my AI video workflow
While building an AI-assisted video workflow for a TikTok commerce creator, I ran into a problem with the storyboards generated by a language model.
- Southeast Asia is hungry for AI — and the money isn't just in subscriptions
Southeast Asia's rapid AI adoption is creating business opportunities beyond paid consumer subscriptions, a partner at Bain & Company, told CNBC.
- What is PHP and why did it spark a global cyber-security scramble?
AI uncovered long-hidden programming flaws, triggering a rare global cyber-security response.
- A Coding Guide to Cactus Whistle: Does a 16.9 MB Speech Model Hold Up Against Whisper Base? Word Error Rate, Speed, Streaming, Timestamps, Keywords, and Speech-to-Tool Calls on a CPU
Explore a comprehensive coding guide benchmarking Cactus Compute's 16.9 MB speech model against Whisper base.
- NVIDIA Nsight Python Tutorial: How to Profile GPU Kernels
Discover NVIDIA Nsight Python, an open-source interface that allows developers to profile GPU kernels, analyze hardware metrics, evaluate cache and clock…
Every story links to its original source. Ranking is by automatic rules, not an editor. Something wrong? Tell us.