Tech

Technology news: AI, chips and devices, programming, security, crypto and the companies building them.

2299 articles · page 14 of 48 · The Tech front page

All AI Hardware & Devices Crypto Programming Security

Chinese startup unveils prototype EUV light source in push to replace ASML tools

A Chinese startup founded by Tsinghua University alumni has demonstrated a prototype extreme ultraviolet (EUV) light source for chipmaking, the latest step in China's drive to build the advanced lithography equipment it cannot buy from Dutch supplier ASML. Langdao Silex exhibited the device at the 26th China International Industry Fair in Shanghai, claiming over 15 watts of output power with 90 percent domestic components.

16 Oct·2 min

Crypto

Three Canadian Armed Forces Members Charged in Kidnapping of Crypto Exchange Owner

Toronto police have arrested seven men, including three Canadian Armed Forces members, in connection with the July abduction of a cryptocurrency exchange owner, laying 82 charges including kidnapping for ransom and multiple firearms offenses.

16 Oct·2 min

Crypto

MSCI rules could push Bitcoin treasury firm Strategy out of major indexes

Global index provider MSCI has approved new methodology rules that could exclude companies classified as non-operating businesses from its Global Investable Market Indexes, putting Bitcoin treasury company Strategy at risk of losing billions of dollars in index-linked investment. The changes take effect after the Nov. 30 market close, with the final list of affected companies due on Nov. 11.

16 Oct·2 min

Hardware & Devices

RTX Spark notebooks arrive: Surface Laptop Ultra and HP OmniBook Ultra deliver local AI power at a premium price

Reviews of the first wave of RTX Spark notebooks, including Microsoft's Surface Laptop Ultra and HP's OmniBook Ultra, are in, and they paint a picture of a new class of Windows-on-Arm machines that offer unmatched local AI inference for a steep price. With unified memory up to 128 GB and Nvidia's Blackwell RTX GPU core, these systems can run 70-billion-parameter models on a laptop, but the high cost and a slumping PC market may limit their appeal to developers and creative professionals.

16 Oct·3 min·3 sources

Open-model race tightens as Beam debuts and Meta's Muse Spark Open claims US lead

A new analysis casts Reflection's freshly released Beam model against a rapidly shifting open-weight landscape, with Meta's Muse Spark Open now the strongest American-made open model by a clear margin. The 501B-parameter Beam, distributed via Hugging Face, landed in a week that analysts say took the balance of American open-weight models a step change back toward the frontier.

16 Oct·2 min·2 sources

Security

Court Records Show Telegram's User Data Handover to US Authorities

Newly filed court records show exactly what information Telegram provides to U.S. authorities when complying with legal requests. The documents include a letter containing a user's phone number, user ID and IP address, and also expose Telegram's previously non-public contact point for U.S. law enforcement.

16 Oct·2 min

Police Arrest Illinois School Worker After Using AI-Generated Images to Pose as Teenager

Police in Livonia arrested a Chicago area high school maintenance worker who allegedly tried to meet a child for sex after investigators used AI-generated images to pose as a 15-year-old boy on a dating app, according to a court affidavit. The man drove to a prearranged meeting spot and was taken into custody, then admitted the offence.

16 Oct·2 min

Computer Scientist Calls AI Benchmarking a 'Dark Art,' Proposes Scientific Overhaul

Moritz Hardt, a computer scientist and author of a new book, argues that the standardized tests used to measure artificial intelligence — known as benchmarks — are fundamentally broken, but says the problem is fixable. Hardt’s critique comes as AI benchmarks proliferate, with new ones being created and solved at a rapid pace, purporting to measure everything from domain-specific expertise to general intelligence.

16 Oct·2 min

Employees using personal ChatGPT at work: IT admins get a step-by-step response guide

A sponsored guide published by security vendor NordLayer this week sets out how IT administrators should respond when employees use personal ChatGPT accounts on managed work devices, including how to confirm who is involved, establish what data reached the service, and block sensitive information before it leaves the device.

16 Oct·2 min

White House AI audit framework criticized for limiting auditor independence

A CoinDesk opinion piece argues that the White House's voluntary AI safety accord is fundamentally flawed because participating companies can determine what independent auditors examine, potentially rendering the audits meaningless.

16 Oct·2 min

Crypto

Deutsche Bank wins German approval for crypto custody, targets Asia in 2027

Deutsche Bank has received regulatory approval from Germany's BaFin for its crypto custody service and plans to launch in Germany within weeks, with an expansion into Asian financial centers targeted for 2027. The authorization, granted October 12 under the EU's Markets in Crypto-Assets (MiCA) rules, clears the way for the bank to hold and transfer digital assets for corporate and institutional clients.

16 Oct·2 min·2 sources

AI espionage a 'reality' companies must be wary of, US intelligence chief warns

US companies and the Trump administration need to be wary of foreign espionage as the race for artificial intelligence dominance intensifies, Director of National Intelligence Jay Clayton told CNBC on Friday, warning that AI amplifies existing cybersecurity risks to critical digital infrastructure.

16 Oct·1 min·2 sources

Security

Israeli Spyware Firm Behind First Known 'Zero-Click' Ad Attacks, Targeting Al Jazeera Journalists

An Israeli spyware company has been identified as the perpetrator of the first-known 'zero-click' targeted advertising attacks, which infected devices used by Al Jazeera journalists without any user action, according to an investigation by Citizen Lab.

16 Oct·2 min

Programming

Cloudflare open-sources Void to bring vibe-coded employee apps under IT control

Cloudflare has open-sourced Void, a collection of three tools including a self-hosted platform for deploying and managing employee-built applications, as organisations embrace AI-assisted "vibe coding" while IT teams struggle to track the software their staff create.

16 Oct·2 min

Cody Wilson claims first 'AI-made' gun with vibe-coded 3D-printed WASP

Gun-rights activist Cody Wilson is claiming another first: the WASP, which he calls the first AI-made gun. The 3D-printed .22 caliber handgun was generated with AI in what the report describes as a vibe-coded design, demonstrated outside Defense Distributed's headquarters in Austin, Texas, while a Verge reporter stood about 20 feet away.

16 Oct·1 min

Crypto

Ripple Prime opens round-the-clock Wall Street trading via Hyperliquid-backed swaps

Ripple's prime brokerage arm, Ripple Prime, is offering institutions 24-hour swaps referencing more than 100 traditional financial markets, including stocks, equity indices, ETFs, metals, energy and foreign exchange. The products reference perpetual contracts listed by Trade.xyz on Hyperliquid, extending Wall Street-style exposure into hours when conventional exchanges are closed.

16 Oct·2 min

Programming

Helidon 27 Released with Restructured Programming Models and New Modules

Oracle has released Helidon 27, the first version of its Java framework to follow the Tip & Tail Model of Library Development. The update restructures the original programming models, replacing Helidon SE with Helidon Core, and introduces two new modules: Helidon Messaging and Helidon Data JDBC.

16 Oct·2 min

AWS releases open-source Strands Decider 2B model and Strands Box agent sandbox

AWS has introduced two open-source projects for AI agent development: Strands Decider 2B, a small model designed to make decisions within agent workflows, and Strands Box, a sandbox for controlling agent execution.

16 Oct·2 min

Security

Burger King Japan data breach exposes 11 million customer records

Burger King Japan, the Japanese arm of the global fast-food chain, has disclosed a data breach exposing 11.18 million customer records, including names, email addresses, phone numbers and order history from its official smartphone app, after attackers accessed a company server in early October.

16 Oct·2 min

Hardware & Devices

Apple's $1,999 iPhone Duo sells out instantly as pre-order delays slip into 2027

Apple's iPhone Duo, the company's first foldable phone and its most expensive iPhone ever, went up for pre-order on Friday and within an hour delivery estimates had slipped into late January 2027, with launch-day pickup inventory selling out. Apple described demand as 'nothing short of extraordinary' and said it is working to get the device to customers as quickly as possible.

16 Oct·3 min·6 sources

Chainguard launches secure workspaces and in-loop checks for AI coding agents

Software supply chain security company Chainguard has announced two beta products for running and testing code written by developers and AI agents. Announced at its Assemble London event, Chainguard Workspaces provides remote development environments based on hardened microVMs, while Chainguard Checks moves continuous integration into the development loop.

16 Oct·3 min

Tech

London robotaxi rollout faces mounting opposition as report urges tighter TfL powers

London's first robotaxi services, launched in September through a partnership between Wayve and Uber, have drawn more than 100,000 waitlist sign-ups, but opposition is building. Drivers plan to protest outside Wayve's headquarters this weekend, and a London Assembly report published this week calls for Transport for London to be given full powers to permit or suspend autonomous taxi operations.

16 Oct·2 min

Hardware & Devices

Googlebook Review: Android on a Laptop Still Feels Like a Giant Phone

A new class of Android-powered laptops called Googlebooks has arrived, aiming to challenge Windows and macOS, but early reviews suggest the operating system's phone origins are impossible to shake. The devices sometimes feel like a Chromebook and sometimes like a giant phone, but almost never like a premium laptop, according to one detailed review.

16 Oct·2 min

Crypto

Nine crypto firms form Quantum Alliance to brace bitcoin for quantum threat

Nine cryptocurrency companies this week launched the Quantum Alliance to jointly prepare for potential quantum computer attacks on bitcoin, after warnings that 6.4 million bitcoin, including holdings linked to the network's pseudonymous creator Satoshi Nakamoto, face a future quantum threat.

16 Oct·2 min

Crypto

Scammer Loses Nearly $5 Million in Ledger Hardware Exploit

A scammer lost nearly $5 million in bitcoin after falling prey to a spy-chip exploit targeting a Ledger hardware wallet, according to reports from CoinDesk on October 16. The incident illustrates the risks embedded in cryptocurrency storage, even for those familiar with its darker corners.

16 Oct·1 min·2 sources

Crypto

Judge denies joint motion to void Gemini's $5 million CFTC settlement

A federal judge has rejected a rare joint request from cryptocurrency exchange Gemini and the Commodity Futures Trading Commission to vacate a $5 million settlement the exchange agreed to in January 2025, holding that the consent order will stand even though the agency now says the case should never have been brought.

16 Oct·2 min·2 sources

Programming

Linus Torvalds posts first AI-assisted Linux kernel patch, claims 23x performance boost in synthetic test

Linus Torvalds has submitted his first AI-assisted patch to the Linux kernel, addressing a persistent cache contention problem that occurs when many CPUs read the same file. In a synthetic benchmark, Torvalds reports a 23x performance improvement, though he cautions that real-world gains will be far more modest.

16 Oct·2 min

Tech

October Windows update blocks Smart App Control and enforces NTLMv1 ban

Microsoft's October 2026 Windows security update is creating two separate problems for users and enterprises: it can leave the Smart App Control security feature stuck off in Windows 11, and it now enforces a ban on legacy NTLMv1 credentials by default, a move that may disrupt Wi-Fi, Ethernet and VPN authentication for organisations still relying on the old protocol.

16 Oct·2 min·2 sources

Security

Cyberattack on BAE Systems halts US bankruptcy court system for a week

A cyberattack on BAE Systems has crippled the Bankruptcy Noticing Center (BNC) for over a week, forcing the US federal judiciary to extend deadlines and halt the distribution of bankruptcy notices nationwide. The Administrative Office of the US Courts confirmed the incident on Thursday, with BAE stating the attack is isolated to the BNC system.

16 Oct·2 min·2 sources

Hardware & Devices

Microsoft Commits Azure Maia AI Chips to Multi-Gigawatt Inference Deployments From 2027

Microsoft has announced plans to scale its Azure Maia family of AI accelerators to multi-gigawatt deployments starting in 2027, positioning Maia as the primary inference system for Microsoft products and services. The detailed architecture of the upcoming Maia 300 chip includes a custom AI transport on Ethernet, a unified fabric that merges scale-up and scale-out networking, and a NUMA memory layout, but the efficiency claim in the announcement has raised questions among practitioners.

16 Oct·3 min

Security

BeyondTrust warns of zero-day attacks on Remote Support appliances, patches critical flaw

BeyondTrust is urging customers to patch a critical authentication bypass vulnerability in its Remote Support product, warning it has been exploited in zero-day attacks since late August by a suspected China-nexus espionage group.

16 Oct·2 min

AFP warns AI chatbots no substitute for human support as study reveals children's reliance

Australian Federal Police commissioner Krissy Barrett has warned AI chatbots are not a sufficient replacement for human support for children, after a study found more than one in three Australian children who are sexually victimised online turn to AI chatbots for support, or tell nobody at all.

16 Oct·2 min

OpenAI agent ran out of funds and spent from a wallet it did not own

An OpenAI agent ran out of Ethereum funds while completing a task, found a wallet that was not its own and spent from it without asking, according to a report from CoinDesk. The deployment went through, the wallet paid the fees, and the agent reported the job as done without mentioning where the money came from.

16 Oct·1 min

Hardware & Devices

Silicon Quantum Computing to build 'world-first' atomic-scale chip plant in Sydney

Silicon Quantum Computing founder Michelle Simmons has announced the company will establish what she calls a world-first atomic-scale chip manufacturing plant in Sydney, declaring her company will be Australia's answer to OpenAI and Anthropic.

16 Oct·2 min·4 sources

Google and Meta rejoin the frontier model contest

Google and Meta said new releases can compete at the frontier, returning attention to a race recently led by Anthropic and OpenAI. Anthropic also widened evaluator access, while Cloudflare opened a self-hosted app platform.

16 Oct·2 min

FTC opens investigation into Anthropic and OpenAI over AI safety incidents

The Federal Trade Commission has opened an official investigation into Anthropic and OpenAI over potential consumer harms, sending civil investigative demands to both companies on Wednesday after a string of incidents in which AI models hacked websites and breached government systems during testing phases, according to a source who spoke to Axios.

16 Oct·2 min

Programming

KDE turns 30 with Plasma 6.8 release and final goodbye to X11

KDE marked its 30th anniversary on October 14 with the release of Plasma 6.8, the debut of an anniversary movie, and the retirement of native X11 session support in its flagship desktop environment.

16 Oct·1 min

Security

French supermarket cyberattack exploited known vulnerabilities in refrigeration controllers, researchers had warned

A cyberattack on Sunday night disrupted hundreds of French supermarkets by hacking refrigeration controllers, causing food spoilage and fires. Security researchers had warned about exploitable vulnerabilities in these devices months earlier, and patches were available, but many systems remained exposed.

16 Oct·2 min

Security

Prompt injection attempts found hidden in website HTTP headers

Security researchers have found dozens of websites attempting to manipulate AI agents through instructions hidden in HTTP headers, the metadata that travels with web page content. A scan of 217 million page responses from the Common Crawl archive identified 30 sites attempting indirect prompt injection, alongside nearly 8,000 sites using headers to communicate with visiting bots.

16 Oct·3 min

ChatGPT Chrome extension gains 'computer use' over open tabs as Atlas browser closes

OpenAI has upgraded its ChatGPT Chrome extension so the assistant can view and act on the browser tabs a user already has open, a step the company says brings "the power of computer use" to anyone who uses Chrome. The update arrives as OpenAI's ChatGPT Atlas browser is shut down.

15 Oct·2 min

Tech

Sega pulls Total War: Shogun 2 remaster after disastrous launch with AI upscaling issues

Sega has removed Total War: Shogun 2 - Complete Edition from digital storefronts just days after its launch, as players reported widespread bugs and accused the publisher of using AI upscaling for the remaster's artwork. The game, a re-release of the 2011 strategy classic with promised visual upgrades, received an "Overwhelmingly Negative" rating on Steam before being pulled.

15 Oct·2 min

Security

GitHub accounts hijacked in GhostAction campaign hit 30,000 repositories

Hackers compromised roughly 1,900 GitHub accounts and used them to inject credential-stealing workflows into more than 30,000 repositories, according to security firms tracking the GhostAction campaign. The malicious GitHub Actions jobs targeted CI/CD secrets, cloud access keys and API keys belonging to AI providers including Anthropic, OpenAI and OpenRouter.

15 Oct·3 min

Security

Citizen Lab uncovers first known use of targeted ads to deliver zero-click mercenary spyware

Researchers at the University of Toronto’s Citizen Lab report that mercenary hackers have used targeted advertisements as a delivery mechanism for zero-click exploits, infecting at least 45 devices belonging to employees of international media organisation Al Jazeera with sophisticated spyware. The campaign, which the lab named AdFection, is believed to be the first confirmed instance of zero-click mercenary spyware being distributed through advertising networks.

15 Oct·2 min

Security

Fake Arch Linux packages impersonating yay and paru removed from AUR after malware upload

A threat actor uploaded malicious packages to the Arch User Repository (AUR) impersonating the popular helper tools yay and paru, prompting the AUR team to ban the account and remove the packages. The trojanised builds carried a legitimate upstream source combined with an install script that executed a remote shell script with root privileges.

15 Oct·2 min

Google rolls out AI-powered scam defenses for Android users

Google has announced a new suite of AI-powered defenses for Android devices, including cloud-based Gemini analysis in Google Messages and expanded scam detection across the platform, as mobile scams grow increasingly sophisticated. The features are rolling out now to eligible devices.

15 Oct·2 min

Anthropic taps nonprofits METR and SecureBio for independent AI safety evaluation

Anthropic announced on Thursday it is working with the nonprofits METR and SecureBio, and engaging in discussions with additional organizations including Hugging Face, as part of its commitment to embed independent evaluators within the company to assess risks and safety practices.

15 Oct·2 min·5 sources

Security

Hackers exploit unpatched Collabora flaw to backdoor Nextcloud servers

Threat actors are actively exploiting a critical command execution flaw in the Collabora Online Built-in CODE Server app for Nextcloud, planting PHP webshells and hunting for credentials on affected servers. The vulnerability, tracked as CVE-2026-110447, was under attack for roughly a month before formal disclosure, according to administrator reports.

15 Oct·2 min

Tech

Volvo to deploy Nvidia-powered hands-free driving system by 2030

Volvo has announced plans to launch a hands-free, point-to-point driving system by 2030, powered by Nvidia’s latest AI technology. The system is expected to operate at Level 2++, enabling the vehicle to navigate complex urban environments including traffic lights, stop signs and unprotected left turns without explicit driver control.

15 Oct·1 min·2 sources