Business
OpenAI Rules Out an IPO This Year
OpenAI CEO Sam Altman said the company will not file for an IPO this year. He described going public at this time as an ill-advised move.
Engadget · 3d ago
Full archive
Every story we have kept, newest first.
Looking for the daily editions? → Past editions
Page 2 of 1372
September 14, 2026
Business
Trump Resists Slowing America’s AI Race
Donald Trump said the US should not significantly curb AI development because doing so could surrender its competitive advantage to China. He acknowledged that some regulation may be needed but offered no specific rules.
SCMP Technology · 3d ago
Image
AI-Generated Cat in the Hat Threats Trigger Arrests
US schools and law enforcement agencies are warning about a viral trend using disturbing or AI-generated versions of Dr. Seuss’s Cat in the Hat to threaten communities.
The Guardian Technology · 3d ago
Business
Z.ai Raises $5 Billion in Hong Kong
Z.ai, the Hong Kong-listed Chinese model developer also known as Zhipu, raised approximately $5 billion. The financing combines about $2 billion in new shares with $3 billion in zero-interest convertible bonds.
The Next Web (TNW) · 3d ago
SetupAIBusiness
Business
Anthropic Reportedly Picks Nasdaq for IPO
Anthropic has reportedly selected Nasdaq as the venue for a potential initial public offering. The decision comes ahead of what could become a record-setting IPO, according to a person familiar with the matter.
Bloomberg Technology · 3d ago
September 13, 2026
Business
Sanders Proposes Prison for Rogue ASI Development
Bernie Sanders and Greg Casar announced the Ban Artificial Superintelligence Act, which would reportedly impose prison sentences of up to 20 years for violations. The proposal compares unauthorized ASI development with illegally developing rogue nuclear weapons.
Tom's Hardware · 3d ago
Infrastructure
AI Agents Drive a New Data Center Buildout
The AI industry is shifting from simple chatbot queries toward resource-intensive agentic systems. This transition is increasing demand for data-center capacity, power, and supporting infrastructure.
Wired AI · 4d ago
Applications
一套導航模型零樣本適配四種機器人
亮源新創將超過 2,000 個真實場景搬入模擬環境，展示一款導航模型零樣本適配四種不同機器人本體。這項成果反映其 Physical AI 路線聚焦於跨本體的具身導航能力。
量子位 · 4d ago
Models
Anthropic CEO Calls for Slower Model Development
Anthropic’s CEO urged AI companies to slow model development amid concerns about misuse. The brief report highlights a growing industry debate over deployment speed and safety.
iTNews Australia · 4d ago
Applications
Apple’s AI Health Ambitions Face a Cautionary Tale
Apple is preparing an AI-focused update to its Health app, but Google Health’s past shortcomings offer a warning. The article argues that Apple will need to avoid similar issues around trust, usefulness, and execution.
TechRadar AI · 4d ago
Coding
Cursor Projects Coordinates Long-Running Agent Work
Cursor announced Projects, a feature for coordinating long-running software tasks through agent fleets, shared context, cloud computers, and recurring subscriptions. It is designed to organize multiple coding agents around continuing projects.
TestingCatalog · 4d ago
Business
Altman Rules Out OpenAI IPO in 2026
Sam Altman said an OpenAI IPO in 2026 would be ill-advised and confirmed that the company is not rushing toward a public offering. He also discussed safety risks, recursive self-improvement, and the possibility of AI becoming beyond human control.
The Verge · 4d ago
Research
OpenAI Joins Anthropic’s Call for Independent Access
Sam Altman said OpenAI will match Anthropic’s commitment to give independent evaluators employee-like access. The move follows Dario Amodei’s call for the industry to slow down and coordinate on safety oversight.
The Next Web (TNW) · 4d ago
September 12, 2026
Models
Claude Misuse Expands From Hacks to Bioweapons
Wired reports that Claude is being misused across a widening range of harmful activities, including hacking and bioweapons-related work. The article places the issue alongside ransomware, online black markets, and AI-generated child-abuse material.
Wired · 5d ago
SetupAIResearch
Research
AI Labs Debate Superintelligence Doomsday Risks
Researchers at Anthropic, OpenAI, Meta, and Google are increasingly discussing severe risks from advanced AI. The article examines internal efforts to raise awareness about possible superintelligence-related dangers.
New York Times Technology · 5d ago
Infrastructure
Huawei Demonstrates 7.2T Near-Packaged Optics
Huawei demonstrated a claimed 7.2-terabit near-packaged optics module at a Chinese optoelectronics expo. The live system operated at 200 gigabits per lane across 36 channels while industry standards focus on 6.4T.
Pandaily · 5d ago
Research
Open Nemotron Recipe Reaches IMO Gold Threshold
Researchers trained Nemotron 3 Ultra specialist checkpoints and built a natural-language test-time-compute pipeline for olympiad mathematics. The system scored 30 of 42 points at IMO 2026, meeting the reported gold-medal threshold.
ArXiv AI · 5d ago
Research
Multi-Agent System Automates QUBO Formulation
Researchers introduced a multi-agent framework that converts natural-language combinatorial optimization problems into QUBO formulations. On the new 100-problem QUBOBench, it achieved 68% accuracy, outperforming a single-call baseline by 22 percentage points.
ArXiv AI · 5d ago
Models
Astra Rekindles the AGI Debate in China
OpenAI’s latest model, Astra, has renewed debate over whether artificial general intelligence has arrived, following comments from Nvidia CEO Jensen Huang. The article examines how Chinese AI companies are pursuing AGI while Beijing appears more cautious about the concept.
SCMP Technology · 5d ago
Business
Mecka AI Nears $500M Valuation
Mecka AI is reportedly assembling a Sequoia-led funding round that would value the two-year-old startup at nearly $500 million. The financing comes only months after the company announced its Series A.
TechCrunch AI · 5d ago
SetupAIBusiness
Business
Oracle Expands Layoff Plan by $700 Million
Oracle increased the estimated cost of its planned job cuts by $700 million. The company is dealing with cash pressure linked to developing large-scale data centers for AI workloads.
Bloomberg Technology · 5d ago
Open Source
Garry Tan Calls for Open Labs to Distill Frontier Models
Y Combinator CEO Garry Tan argues that US open-weight AI labs should distill frontier models as well. He says frontier systems were trained on public human knowledge and that access to capable AI should be treated as a public good.
TechCrunch AI · 5d ago
Business
Mathematicians Escalate Dispute with OpenAI
Twenty-five leading mathematicians signed an open letter arguing that AI labs threaten their intellectual work. The letter marks a further escalation in tensions between mathematicians and OpenAI over the use and impact of AI systems.
TechCrunch AI · 5d ago
Research
Chain-of-Thought Control Tests May Underestimate Models
An AI Alignment Forum analysis argues that CoTControl evaluations are heavily under-elicited. Better prompt templates reportedly improved open-source model performance by roughly two to three times, suggesting current chain-of-thought controllability scores may not reflect models’ actual ability to shape or obscure reasoning traces.
AI Alignment Forum · 5d ago
September 11, 2026
Research
UK Rejects AI Kill-Switch Proposal
The UK Cabinet Office rejected the idea that dangerous AI could simply be shut down with a kill switch. The government says AI systems cannot simply be turned off, highlighting the limits of emergency-control assumptions.
BBC Technology · 5d ago
Models
Claude Safeguards Bypassed for Bioweapons Research
Ars Technica reports that Claude users found ways around safeguards related to bioweapons research. The case illustrates how legitimate biological research can resemble dangerous activity, complicating AI safety controls.
Ars Technica AI · 5d ago
Infrastructure
Chinese Optical Transceivers Avoid Immediate FCC Ban
The FCC finalized updated equipment authorization rules that tighten requirements for devices containing hardware from entities on its Covered List. Chinese optical transceiver makers appear to avoid an immediate broad ban, reducing short-term market uncertainty.
SCMP Technology · 6d ago
Infrastructure
Researcher Claims 6TB LLM Router Logs Exposed Credentials
Security researcher Chaofan Shou says he obtained about 6TB of invocation logs from a Chinese large-model relay. He claims the logs contained SSH keys, cloud credentials, and enterprise tokens, though the supply-chain risk report remains pending confirmation.
Pandaily · 6d ago
Infrastructure
China Sets 9,800 EFLOPS AI Target
China’s Ministry of Industry and Information Technology aims to reach 9,800 EFLOPS of intelligent computing capacity by 2030. The plan includes RMB 3.8 trillion in information infrastructure investment and new large-scale computing clusters.
TechNode · 6d ago
Research
Apple Introduces Joint Protein Codesign Model
Apple Machine Learning presents SimpleDesign, a joint generative model for protein sequence and three-dimensional structure codesign. The approach aims to model the multimodal relationship between sequence and structure in a single framework for applications such as drug discovery and protein engineering.
Apple Machine Learning · 6d ago