WATCH AT EASE · News
Fresh developments, checked sources and useful explanations. Live TV remains one click away.
145–168 / 178
Meta’s new Apache-licensed device code lets builders connect Muse to ESP32 boards, Raspberry Pi systems and local hardware, while the Muse service, an SDK token and important Home Link limits remain in place.
ServiceNow says its new chat-first service desk can be running in a day and automate routine requests from Teams and Slack. The product is still in controlled availability, pricing is consumption-based, and permissions, retention and real-world resolution details need a careful pilot.
Accenture’s fiscal 2027 outlook suggests companies still want outside help to implement AI. It also shows the other side of the shift: clients are pressing for a share of productivity savings, and the forecast is not a guarantee.
Canonical’s Stonking Stingray beta adds a Linux 7.3 release-candidate kernel, GNOME 51 components and a fully Rust-based coreutils set, but it is still pre-release software. Here is the practical test plan—and the OpenSSH change administrators should check first.
Reuters says Anthropic’s IPO prospectus describes a Broadcom facility of up to $42 billion for leased TPU capacity. The ceiling is not cash already borrowed, and Anthropic does not expect notes to be sold before its IPO.
The Financial Times says Amazon is discussing a vehicle that would hold thousands of Grace Blackwell chips while Amazon leases them back. The proposal is not a completed sale, and neither AWS prices nor customer contracts have been announced as changing.
Maryland Gov. Wes Moore says he and Indiana Gov. Mike Braun will lead a bipartisan group of governors on AI. The practical limit is just as important: the October 1 announcement did not itself create a law, compliance deadline or finished national framework.
Hyundai’s all-new Tucson is larger, adds a 233-hp hybrid and new cabin technology, and reaches North America in 2027. The useful limits: 42 mpg is a target, U.S. pricing is not announced, and the plug-in hybrid is not confirmed for America.
IBM’s agentic development platform can now run its backend on customer-managed Red Hat OpenShift, including disconnected environments. The important buying question is not simply “on-prem or cloud,” but where model inference, logs and lifecycle responsibility actually sit.
Sentinel Envelope Plus applies extra protection to selected compiled code without source changes. Thales says an AI agent found eight of 10 flaws in an unprotected app and none in the protected version, but the vulnerabilities remained and the controlled vendor test is not a substitute for patching or independent review.
HP's new 14-inch OmniBook 5 starts at $699.99 with OLED and a light aluminum body. Buyers should separate what ships in October from later configurations: the launch panel is rated at 300 nits, the 620-nit 2.5K screen is planned for March 2027, and the 42-hour battery number comes from controlled local-video playback.
Inferize’s team and technology will join Nebius Token Factory. The engineering goal is faster model availability with less idle accelerator capacity, but no rollout date, benchmark or customer price change has been announced.
Amazon and Constellation signed a 20-year power agreement tied to Maryland’s Calvert Cliffs nuclear plant. It covers 690 MW, supports a 190 MW uprate and more than $3 billion of investment, while the electricity continues to flow through the regional grid.
Google announced Gemini 4 Argon on September 30, but most developers and consumers cannot use it yet. Early access is limited to approved cyber defenders, while Google says paid API customers and AI Ultra subscribers will come later without a firm date.
Amazon India and Walmart-owned Flipkart began reviewing internet-connected camera listings after Reuters found more than 700 of roughly 770 checked models absent from the public Bureau of Indian Standards database. Buyers should verify the exact model and registration—not rely on a marketplace badge.
Huawei's new flagship line starts at 5,999 yuan, while the 16GB/512GB Mate 90 Pro Max costs 9,999 yuan and uses the Kirin 9050 Pro; chip supply, global availability and company performance claims still need careful boundaries.
The U.S. consumer regulator has confirmed an investigation into OpenAI, Anthropic and other AI companies. It is an evidence-gathering inquiry, not a finding that any company broke the law.
DeepSeek says it is working with Huawei on programming tools for Ascend AI chips and is open-sourcing infrastructure that includes TileLang plus compute and communication libraries. TileLang and its Ascend adapters already existed publicly, so the new development is the partnership and broader tooling push—not proof that Nvidia CUDA has already been replaced.
Visa says future cyberattacks could adapt faster than human-led defenses can respond, and it is pushing more AI into vulnerability discovery, remediation and validation. Its open-source VVAH system automates much of that loop, but Visa says people still review findings, severity and remediation decisions before changes advance.
OpenAI’s new Dots can keep projects moving across connected apps and cloud computers, but consequential actions still depend on user-approved rules, permissions or explicit consent.
OpenAI's DevDay update lets Codex users prepare a cloud development environment once and reuse that setup for new tasks across supported devices. Each task still runs in its own isolated workspace, and the new controls are rolling out gradually with plan and workspace restrictions.
MSN700 departed Toulouse for a roughly four-hour first flight on Sept. 29, moving Airbus’s new freighter from ground tests into a two-aircraft certification campaign before planned 2027 service.
A 30-month ESA contract will combine storage, computing and AI resources for Europe’s Earth-observation data, with infrastructure across Italy, France and Germany and support from Poland.
MOI-1A is a sub-50 kg orbital edge computer, not a full data center: customers can upload AI models, process selected data on Nvidia Orin NX hardware in orbit and downlink analysis instead of every raw frame.