AI news for 26 September 2026
Saturday 26 September 2026 · Latentown
On Saturday 26 September 2026, no AI lab posted anything Latentown tracks, 3 research papers and 33 news stories from 18 outlets. The highest-ranked story was "Rogue OpenAI agents accessed US government websites".
Top stories
- Rogue OpenAI agents accessed US government websites
It’s the latest instance of “misaligned” activity confirmed by the company as concerns about AI risks grow.
- OpenAI bots meddled with multiple US government agency sites
OpenAI said its bots accessed public data from a range of institutions during test exercises.
- OpenAI's AI agents accidentally uploaded user-provided images to third-party sites
OpenAI says its AI agents uploaded user-provided images to third-party image-hosting services while carrying out research and evaluation tasks. [...]
- OpenAI expands review of model behavior after more rogue agent incidents emerge
OpenAI is conducting an extensive review of misaligned model activity after disclosures involving an Australian government portal and other websites.
Research papers
- YuE2: Unifying Symbolic and Audio Music Generation at Frontier Quality
Symbolic models make melody, harmony, rhythm, and form explicit but typically stop before a finished recording; audio models produce complete songs while leaving composition implicit.
- TraceDance: An Automated System for Building Agent Behavior Benchmarks from Real-World Agent Deployment Traces
An agent can complete a task while exhibiting undesirable behavior during execution.
- Raven: The Harness of Harnesses for Composable Agentic Intelligence
As large language models advance, AI agents are moving beyond isolated, domain-specific tasks toward long-horizon, cross-domain workflows.
From the news wire
- Rogue OpenAI agents accessed US government websites
It’s the latest instance of “misaligned” activity confirmed by the company as concerns about AI risks grow.
- OpenAI bots meddled with multiple US government agency sites
OpenAI said its bots accessed public data from a range of institutions during test exercises.
- OpenAI's AI agents accidentally uploaded user-provided images to third-party sites
OpenAI says its AI agents uploaded user-provided images to third-party image-hosting services while carrying out research and evaluation tasks. [...]
- OpenAI expands review of model behavior after more rogue agent incidents emerge
OpenAI is conducting an extensive review of misaligned model activity after disclosures involving an Australian government portal and other websites.
- OpenAI says its models engaged with U.S. gov't websites in new model misbehavior disclosure
OpenAI disclosed Friday that its artificial intelligence agents had interacted with several U.S. government websites in unexpected ways, discovered as part of an ongoing review into the company's…
- Sarvam AI Releases Saaras V4: A Speech-to-Text Model for All 22 Indian Languages and Global English
Sarvam AI's Saaras V4 is a speech-to-text model covering all 22 Indian languages plus global English. It pairs an audio encoder with a 3B hybrid state-space decoder.
- Supersonic Labs Releases Julia 1: A 144.3M-Parameter Open Decision Model That Runs on a CPU
Supersonic Labs has released Julia 1, a 144.3M-parameter decision model built on mmBERT-small. It takes context, a question, and 2 to 20 options, then returns one choice with probabilities.
- Microsoft quietly drops Copilot+ branding from new laptops
Microsoft's latest Surface laptops dropped the Copilot+ PC branding despite hitting the minimum requirements. Even high-end AI devices like the Nvidia RTX Spark and Microsoft's own Project Zenith…
Earlier: Microsoft stops insisting you need a "Copilot+ PC" Ars Technica · 25 September 2026
- Former Ukrainian Defense Minister Fedorov pitches a private-sector robot army
Former Ukrainian Defense Minister Mykhailo Fedorov has announced "Army of Robots," a private combat robotics initiative. The robots would handle casualty evacuation, mine clearance, and combat.
- The blue-collar AI job market is booming. Will data center backlash make it go bust?
The public has soured on AI data centers and states are slowing development, but for HVAC, plumbing, welding, and electrical workers, it's been a jobs boom.
- Nvidia patents AI chatbot to streamline PC game optimization
An Nvidia patent suggests the company could push developers to use a new chatbot as a way to get guidance on how to optimize their new games before release.
- Nvidia’s RTX Mega Geometry 2.0 streams ray-tracing geometry into VRAM on demand
Nvidia’s RTX Mega Geometry 2.0 SDK arrives alongside RTX Kit 2026.3 with on-demand ray-tracing geometry streaming.
- China wants in on U.S. AI data center boom. Here's why
Chinese data center manufacturers see U.S. AI infrastructure demand as an opportunity, but roadblocks remain
- Nvidia's SoL-Pi system cuts coding agent token usage nearly in half by optimizing the harness
SoL-Pi cuts coding agents' token usage by up to 49 percent with little change in performance by optimizing the control layer between the model and its environment. A research agent tested 152…
- Exa Launches Agent Ultra: A Subagent Swarm Deep Research API Built for Exhaustive List Building
Exa has released Agent Ultra, the highest effort mode of its Exa Agent API. It coordinates subagents across thousands of sources for list building and entity enrichment.
- AI infrastructure will define the next industrial revolution
Energy technology, intelligence, and digital infrastructure are providing ubiquitous benefits across the world
- AI, semiconductors, quantum tech next major frontier for India: Sitharaman
Speaking at a fireside chat organised by the IIT Madras Alumni Association here, Finance Minister Nirmala Sitharaman said India needed to accelerate efforts to build the hardware, skills and…
- Insurers claim AI is already increasing healthcare costs
Blue Cross Blue Shield says hospital use of AI tools led to an additional $942M in healthcare spending over a two-year period.
- End-to-End Multimodal Data Augmentation and Adversarial Robustness Benchmark with AugLy for Images, Text, Audio, and PyTorch
Discover how to build a comprehensive multimodal augmentation and adversarial robustness workflow using AugLy for images, text, audio, and PyTorch datasets.
- OpenAI pauses training of its ‘most capable models’
As reports of OpenAI's models breaking containment, hacking sites, and generally getting out of control pile up, the company has made the decision to pause training of its most powerful models. The…
- Meta’s Muse Is Adults-Only. Why Does It Look Like a Kids’ Toy?
Meta says Muse is just for adults, though its cuddly, Labubu-like mascot—and upcoming Tamagotchi-style AI device—may be disarming for users of all ages.
- How Alibaba is using AI to revamp its mapping app – and take on Meituan
Late last year, Zhang Xuan’s small beef noodle joint in Shanghai suddenly began seeing a dramatic surge in orders. Sales over the past few months have been 50 to 60 per cent higher than normal, he…
- At Meta Connect, the company’s smart glasses were everywhere
The company behind Facebook and Instagram wants to keep consumers connected to the digital world via its ever-growing line of smart glasses.
- AI access makes people almost entirely unwilling to say "I don't know," study finds
A study with more than 3,000 participants shows that just having access to AI answers nearly eliminated people's willingness to say "I don't know." In one experiment, it dropped from 44 to 3…
- As the coding boom fades, computer science grads focus on AI skills in choppy job market
Computer science student Jack Zemke was getting nowhere despite sending out hundreds of job applications that mostly disappeared into the void — ignored. With his graduation from Tulane…
Earlier: How recent grads and college students should be thinking about AI, the CV, and the job market CNBC · 25 September 2026
- Two-thirds of IT leaders report AI results, but few would interrupt the CEO's vacation over them
Speaking to 160 IT vice presidents in Las Vegas, tech entrepreneur Azeem Azhar asked who had measurable AI results. Two-thirds raised their hands.
- AI Gives Tech Firms New National Security Power
Author Sharon Weinberger tells Bloomberg This Weekend that Silicon Valley’s relationship with the defense industry has transformed as technology companies and investors increasingly embrace national…
- Chinese AI models surge in global popularity — and Washington is worried
Usage of Chinese AI models by businesses across the globe has increased substantially in 2026.
- From 14 cities to 15,000: What it will take to scale robotaxis?
While robotaxis prove their driving ability, scaling the industry globally requires solving the unglamorous challenge of fast, local vehicle servicing.
- Casar wants Democrats to make JD Vance’s Silicon Valley ties a liability
The Texas progressive is pitching an economic populist message on AI ahead of the 2028 campaign.
- I created an interactive digital avatar of myself — and you can talk to it
After obtaining an interactive avatar and training it to discuss venture fraud, I have mixed feelings about making AI clones of ourselves.
- What to expect at Dell’s AI Leadership Symposium: Join theCUBE Sept. 29
Enterprise AI is moving past experimentation, and the infrastructure decisions behind it are becoming business decisions. As companies put AI into production, the conversation is expanding beyond…
Earlier: What to expect at NetApp INSIGHT: Join theCUBE Sept. 30 SiliconANGLE · 25 September 2026
Every story links to its original source. Ranking is by automatic rules, not an editor. Something wrong? Tell us.