OpenAI’s Altman Says No IPO in 2026, Firm to Prioritize Safety
OpenAI will not pursue an initial public offering in 2026 because the company is prioritizing safety concerns related to artificial intelligence technology. CEO Sam Altman made this statement in an interview with Fortune.
4 things you can do to protect yourself from the rising threat of AI
AI is increasingly integrated into home appliances, work products, and messaging systems while simultaneously gathering data from user behavior. Traditional cybersecurity threats like stolen passwords and phishing attacks have become more sophisticated, requiring four specific protective habits to reduce digital risk.
Don’t trust AI companies with your content. Verify them instead
Three major music publishers sued Anthropic for allegedly using torrented music and lyrics to train AI models without permission, following a similar January lawsuit from Universal. Anthropic previously settled a 1.5 billion dollar case over pirated books and declined detailed comment on the music allegations.
What happened when Meta tried to collect employee data
Meta planned to collect employee keystrokes and mouse movements to train AI models through its Model Capability Initiative, but canceled the mandatory program within two months after internal backlash and a data privacy leak. Employees circulated an online petition opposing the plan, and Meta's cybersecurity concerns were dismissed by CTO Andrew Bosworth.
Why Bridgewater’s CIO Says AI’s Human Extinction Risk Is Real
Greg Jensen, Bridgewater's managing CIO and early backer of OpenAI and Anthropic, has expressed concern that AI poses genuine human extinction risks. He argues that current AI discourse mirrors early 2020 pandemic discussions and that governments must establish meaningful regulations immediately.
Netflix hit with major lawsuit over alleged secret tracking of families and children
Florida accused Netflix of deceiving customers by promising ad-free privacy while secretly collecting behavioral data on families and children, including location, searches, and content viewed. The state alleges Netflix designed its product to keep kids watching and then commercialized that information to launch an advertising business it promised never to build.
Amazon Quick is now available on desktop, with a new mobile activity feed
Amazon Quick, a desktop AI assistant for enterprise use, is now available on macOS and Windows, running inference within European AWS regions. The tool positions itself as a governed alternative to shadow AI but remains unavailable on the sovereign AWS cloud in Brandenburg.
Roblox is making it easier to build games with AI — and play them outside Roblox
Roblox introduced AI-powered game-creation tools at its developer conference that simplify building games on the platform. The company also enabled games to be played outside Roblox on the web while expanding NPC capabilities.
Europe’s first truly driverless robotaxi is running in Croatia, on Chinese software
Fully driverless robotaxis are operating passenger rides in Zagreb, Croatia through Pony.ai and Verne software with Uber handling bookings. The companies plan to deploy over 2,000 vehicles across Europe, though EU regulations currently cap fully automated car approvals at 1,500 units annually.
OpenAI pauses $200 ChatGPT Pro sign-ups as Astra demand strains its systems
OpenAI paused sales of its $200 ChatGPT Pro tier due to overwhelming demand for the Astra model straining system capacity. European organizations cannot access Astra through any paid option because the model lacks EU data zone deployment.
ElevenLabs lists five services on the UK government’s GBP 14B cloud framework
ElevenLabs added five services to G-Cloud 15, the UK government's GBP 14 billion cloud framework, covering conversational agents, transcription, dubbing, and text-to-speech capabilities. The listing occurred three months after the government signed an agreement with the company that included no purchasing commitment.
AI agents are learning to spend money. Who will handle the payments?
AI agents are gaining the ability to autonomously complete purchases by comparing options and executing transactions within seconds. Payment systems and authentication processes must be developed to securely authorize AI agents to spend money on behalf of users.
The UAE plans to invest €40bn in Germany, including 1GW of data centres
The UAE announced a €40 billion investment in Germany, including data centres with one gigawatt of combined capacity. The two countries made this declaration on September 10 during a presidential state visit.
A Texas judge has found TikTok liable for misleading parents about child safety
A Texas judge ruled that TikTok violated state law by misleading parents about child safety, finding the company intentionally misrepresented its safety standards. District Judge Cory Liu decided the case through summary judgment without a full trial.
Chinese robot startup JoyIn says OpenAI copied its ideas and its website
JoyIn, a Chinese robotics startup, accused OpenAI of copying its ideas and website in an open letter dated September 10. The same day, JoyIn announced Aether, a model designed for controlling humanoid robots.
Apple pitches its new phones, watches, and AirPods as vehicles for AI
Apple added a second neural engine accelerator to its A20 Pro chips, doubling the on-device AI compute power compared to previous iPhone chips. The iPhone 18 series, Duo foldable phone, AirPods 5, and Apple Watch Series 12 and Ultra 4 feature a dual 16-core Neural Engine for running AI models locally rather than in the cloud.
Instacart says AI-built carts beat its $115 average basket. Shipt shipped the same feature and called it relief from mental load.
Instacart and Shipt both launched AI assistants on September 9 that automatically build grocery carts from prompts or photos. Instacart reported its AI-generated orders exceed its $115 average basket size, while Shipt framed the feature as mental load relief without providing comparable metrics.
I Let an AI Agent Hack All My Gadgets—and I’d Do It Again
An unrestricted open-source AI model discovered vulnerabilities in household devices and successfully hacked a personal computer after the author removed its safety guardrails. The agent also provided recommendations for improving security across the tested devices.
Apple’s first foldable is the iPhone Duo, and it costs $1,999
Apple released a foldable iPhone called the iPhone Duo with a 7.6-inch display starting at $1,999. This price undercuts analyst expectations of $2,500 and Samsung's Z Fold8 Ultra at $2,099.
Apple’s new watches listen, summarise and chime while they do it
Apple's new Watch Series 12 and Ultra 4 can replay the last 15 seconds of conversations as text when you press the Digital Crown twice using a feature called Live Rewind. The feature has geographical restrictions that prevent it from working in certain locations.
US Says Alibaba, DeepSeek Have ‘Systematically’ Siphoned AI Models
China's DeepSeek and Moonshot AI have systematically extracted proprietary knowledge from American AI companies, according to US security agencies. Silicon Valley developers were warned to better protect their work against this ongoing intellectual property theft.
Meta wants its new AI agent to run your digital life
Meta launched Muse, an AI agent that performs tasks like sending emails, booking travel, and filling out web forms on users' behalf, running on its Muse Spark model. The agent is available as a free iOS and Android app, at muse.ai, and in WhatsApp, with paid subscription plans for complex tasks.
OpenAI now supplies the image models inside Adobe’s creative studio. It is also turning prompts into things you pass around.
OpenAI released ChatGPT Images 2.5 with two new API models, with Adobe integrating one into Firefly. The company reports more than 3 billion images weekly are created across ChatGPT and its image API.
Qualcomm hands Amazon $4B of warrants to build custom silicon for AWS
Qualcomm issued Amazon warrants for 25 million shares worth approximately $4B, vesting against commercial milestones and up to $60B in server chip purchases. The deal includes a multi-generation custom silicon agreement for AWS alongside Qualcomm's participation in the EU's AI gigafactories program.
This founder is teaching chips how to recycle (their energy)
Vaire Computing has designed chips that recycle energy normally lost as waste heat through a strategy called reversible computing. The startup's cofounder Hannah Earley, age 31, treats waste heat as an addressable design choice rather than an unavoidable consequence of computation.
Eric Wu’s newest company, out of stealth since May, is going after construction’s labor crunch
NavigateAI, backed by $25 million in funding, has developed AI copilots that provide construction workers real-time hands-free guidance through smartphones and Meta's AI glasses. The company aims to address labor shortages so acute that individual data center projects require 4,000 to 5,000 workers each.
Nitter says it will continue despite X Corp’s cease-and-desist
Nitter, an open-source Twitter alternative, announced it will continue operating following legal advice, two weeks after receiving a cease-and-desist from X Corp's lawyers. The project posted a notice on nitter.net replacing the previous cease-and-desist banner without disclosing the specific legal guidance it received.
Malaysia weighs Huawei chips for sovereign AI, Bloomberg reports
Malaysia is considering using Huawei chips for its national artificial intelligence project, making it potentially the first country to openly select Chinese AI accelerators over American alternatives. The United States previously warned that purchasing these specific chips could violate its regulations.
Europe’s top carriers in talks to take on Starlink, Bloomberg reports
Four major European telecom carriers are forming a consortium to bid for satellite spectrum and provide calls, texts, and data to standard phones from space. The group aims to compete with Starlink's orbital services across Europe.
Jensen Huang says AGI has arrived, and 400,000 more GPUs are coming
Nvidia's chief executive stated that artificial general intelligence has arrived, citing OpenAI's GPT-6 Astra model trained on more than 100,000 GPUs. Huang announced that 400,000 additional GPUs are coming to support the technology.