Plex is tightening its remote‑streaming policy this week, moving away from free, unrestricted access in order to cover rising infrastructure costs, fund new features such as Common Sense Media integration, a bespoke server‑management app and an open API, and boost subscription revenue after nearing profitability and raising $40 million in 2024; the change may push longtime users who rely on Plex as a free media‑server solution toward alternatives like Jellyfin, while the company hopes the new model will satisfy investors and sustain its growth. Click here to read more.

“Agentic AI” refers to systems that go beyond chat‑based language models by autonomously planning, acting and learning to achieve high‑level goals without step‑by‑step human instruction, a concept highlighted in a new MIT‑Boston Consulting Group report that surveyed 2,000 executives and described these agents as “autonomous teammates” capable of multistep processes; industry leaders such as Amazon’s AWS, OpenAI, Google and Microsoft argue that combining large‑language models with task‑execution capabilities will let agents handle complex workflows—from purchasing and travel booking to managing medical bills or filtering spam—while critics note the term’s recent marketing hype and emphasize the need for clear definitions and safeguards as these agents gain more freedom and responsibility. Click here to read more.

Utilities are projecting that data‑center demand could double or triple within a few years, prompting lawmakers and regulators to question whether those forecasts are based on projects that will actually be built or on speculative proposals that could leave ratepayers footing the bill for unnecessary power plants and grid upgrades; the mid‑Atlantic grid operator PJM and Texas regulators have highlighted concerns about duplicate requests and lack of transparency, while industry groups such as the Data Center Coalition and utilities like PPL argue that many projects are financially committed, leading to new disclosure laws and calls for clearer verification of commercial readiness to avoid over‑building capacity. Click here to read more.

VR and XR headsets remain a niche market, with premium devices like Apple’s Vision Pro and Samsung’s Galaxy XR accounting for only about 5‑6 % of shipments and selling mainly to developers, enterprises and specialized professionals who need large virtual workspaces, medical imaging or training tools, while affordable Meta Quest units dominate consumer volumes but still pale beside the billions of smartphones sold; the industry treats headsets as a transitional platform for future smart‑glass products, prompting big‑tech firms to continue heavy investment despite modest sales and operating losses. Click here to read more.

Analysts predict Valve’s upcoming Steam Machine will likely launch between roughly $800 and $1,100 depending on storage capacity, with some expecting a high‑end $1,000‑plus price that rivals a PS5 Pro and others forecasting a more aggressive $550‑$750 entry point that could serve as a loss‑leader to grow the SteamOS ecosystem; pricing uncertainty stems from volatile RAM and SSD costs, potential tariff impacts that could add $50‑$100 in the U.S., and Valve’s strategic choice between higher margins for a premium niche market or lower margins to attract broader adoption, while the company remains silent on exact figures pending component pricing and supply‑chain considerations. Click here to read more.

Valve is re‑entering the VR arena with the Steam Frame, a standalone headset that weighs just 440 grams—thanks to a 185‑gram modular “core” that houses the processor, displays, tracking and other components—while a dedicated wireless adapter and new Foveated Streaming technology aim to deliver low‑latency, battery‑efficient gaming without any wired PC connection; the 21.6 Wh battery offers variable runtime depending on game settings, and a Gen 4 PCIe expansion port lets developers attach accessories such as monochrome passthrough cameras, opening the door for third‑party customizations in a market now dominated by AR‑focused rivals. Click here to read more.

Leaks of unusually long ChatGPT prompts have been surfacing in Google Search Console (GSC) since September, revealing that user queries—often personal or business‑related—were inadvertently sent to Google when the AI’s web‑search feature was triggered by a buggy prompt box that appended “https://openai.com/index/chatgpt/” to each request. Analysts Jason Packer and Slobodan Manić traced the issue to a glitch that caused ChatGPT to route search‑enabled prompts to Google, which then logged the raw queries in the GSC of sites ranking for the “openai index chatgpt” keywords. OpenAI acknowledged the problem, stating it had fixed a temporary routing glitch affecting a small number of queries, while Google declined comment. The episode raises lingering concerns about privacy safeguards, the scale of exposed prompts, and whether the fix fully stops the leakage of user data to third‑party services.  Click here to read more.

Quantinuum’s new Helios processor expands trapped‑ion quantum hardware from 56 to 96 qubits while preserving high two‑qubit gate fidelity, thanks to a novel loop‑and‑leg architecture that shuttles ions through a central four‑way intersection to bring any pair together for operations. The system’s active‑controlled parachute‑like ion routing reduces traffic jams, and its GPU‑driven real‑time control engine, paired with the updated Guppy Python SDK, enables dynamic error detection, correction and conditional programming. Using Helios, researchers simulated a Fermi‑Hubbard model of superconductivity, achieving accurate results despite modest error rates, and demonstrated the ability to configure the 96 physical qubits as 48 logical qubits via a concatenated error‑correction code. Quantinuum envisions future grid‑based chips that build on Helios’ junction reliability, aiming for larger, faster quantum processors that can tackle problems beyond classical reach. Click here to read more.

Microsoft has rolled out its first home‑grown text‑to‑image model, MAI‑Image‑1, now integrated into Bing Image Creator and the new Copilot Audio Expressions “story mode.” According to AI chief Mustafa Suleyman, the model excels at producing photorealistic food, nature, and artistic lighting scenes while delivering faster results than many larger competitors, allowing users to quickly iterate and export assets to other tools. MAI‑Image‑1 joins DALL‑E 3 and GPT‑4o on Bing’s image platform, marking a shift for Microsoft toward more in‑house AI alongside its existing partnerships with OpenAI and Anthropic. Click here to read more.