News without the noise
Technology news: AI, chips and devices, programming, security, crypto and the companies building them.
2284 articles · page 48 of 48 · The Tech front page
A mysterious new artificial intelligence model named Ox Alpha has emerged, sending certain corners of the internet into a frenzy of speculation about its origins, capabilities, and the identity of its creators. With no official announcement or documentation, the model's sudden appearance has become one of the most talked-about topics in AI communities this week.
Crypto-powered card spending has topped $1 billion in tracked volume, more than tripling year-over-year, as stablecoins increasingly fund everyday purchases such as groceries, rides and subscriptions. At the same time, a Coinbase executive has described the current stage of AI agent payments as the ‘Napster/LimeWire era,’ suggesting that automated agents could become the next billion users for crypto payments.
Waymo, the autonomous-driving subsidiary of Alphabet, is doubling down on custom-designed chips to drive its robotaxi ambitions, according to a recent TechCrunch report. The move underscores a broader industry trend toward vertical integration as companies seek to overcome the technical and economic hurdles of scaling self-driving vehicles.
Flock Safety CEO Garrett Langley has urged critics and supporters to find 'common ground' as his company — which operates a vast network of AI-powered license-plate cameras used by thousands of police departments — confronts a growing wave of public backlash over surveillance, privacy, and potential misuse of its technology.
The U.S. Army has tested an autonomous Ford F-250 truck equipped with a computer-controlled shotgun turret designed to destroy enemy drones, officials announced after a live-fire exercise at Fort Bragg on August 18 as part of the Project Sandhills 2.0 initiative.
A new paper published in the International Journal of Cultural Property warns that mummified human and animal remains—already sought after in a clandestine collectors' market—may pose serious health hazards due to toxic preservation chemicals and long-dormant microbial growth, reviving the scientific basis for the legendary 'mummy's curse.'
The U.S. Federal Communications Commission has retroactively revoked certification for the HoverAir Versa, a 230-gram modular camera that can snap into a self-flying drone, just days after the device raised more than $230,000 on Indiegogo. The FCC stated the authorization was 'granted erroneously' and that the device constitutes a prohibited foreign-made unmanned aircraft system (UAS) critical component, apparently in willful violation of a December 2025 order banning such equipment.
Universities across New South Wales are increasingly embedding artificial intelligence into their teaching, rolling out tools such as AI-generated patient simulations for nursing students, avatar versions of course conveners, and an 'academic support assistant' that fields student queries, according to newly revealed details.
A passenger's laptop caught fire aboard American Airlines Flight 2398 on Saturday, prompting the pilot to report a "cabin fire in the back" before the aircraft landed safely. A passenger was treated at the airport, according to BBC News.
Security researcher Christopher Domas has disclosed a hardware-level vulnerability in modern CPU architectures that allows unprivileged software to access protected memory regions by manipulating DRAM controller translation registers, raising significant concerns for cloud and confidential computing environments.
Bitcoin surged past $75,000 for the first time on Thursday, extending a rally that has seen the cryptocurrency gain more than 40% this year. The milestone, recorded in early Asian trading, was driven by heavy institutional buying and optimism around spot Bitcoin ETF inflows, according to market analysts. 'Bears in pain, right now,' wrote bitcoin analyst James Check, reflecting the overwhelming bullish sentiment.
Nvidia researchers have demonstrated that fine-tuning can enable AI agents to perform complex tasks reliably even when the underlying language model is not particularly strong at the task, suggesting that the system design around a model, rather than the model's inherent capabilities, may be the decisive factor in agent performance.
The Sandbox, a prominent Web3 gaming metaverse platform, has disabled bridging services on the Base and BNB Chain networks after detecting an exploit, with the company confirming that less than 0.01% of the total SAND token supply was impacted. Users are being warned not to trade SAND on these chains, while investigations into the incident are ongoing.
OpenAI, the developer of ChatGPT, has reversed its stance on California's proposed AI safety bill, SB 53, and is now urging state lawmakers to strengthen the legislation rather than weaken it—a move that has surprised observers and reignited debate over how to regulate advanced artificial intelligence.
A new study published August 22, 2026, has found that leading artificial intelligence labs have few publicly documented plans for containing a rogue AI model, sparking concerns about preparedness as systems grow more capable and unpredictable.
Anthropic has announced it is now running its most powerful AI model, Claude Mythos 5, through its Claude Security vulnerability scanner, offering $35 million in credits to open source projects. The move marks a significant deployment of a model the company previously held back from public release, now tasked with scanning codebases for flaws and integrating with partner products protecting critical infrastructure.
In a landmark cyberattack, hackers linked to the Iranian regime have successfully shut down a power plant in the United Kingdom, marking the first time such a facility has been taken offline by state-affiliated actors, according to security officials and reports from The Financial Review.
Harvard Business School’s HBS Foundry program has launched a $699 online startup bootcamp that employs AI-powered avatars of its instructors to provide feedback during practice pitches and board meetings, blending the school’s prestige with artificial intelligence to reach a wider audience of aspiring founders at a fraction of the cost of its executive courses.
Thomson Reuters is building its own language model called 'Thomson,' based on Alibaba's Qwen architecture, investing approximately $40 million over two years to own its AI rather than rent from OpenAI or Anthropic. The move comes as Inherent, a British AI lab founded by DeepMind alumni, releases Faraday, an AI agent that claims to outperform rivals at replicating scientific research, signaling a broader trend toward domain-specific AI in enterprise and science.
Citing a sharp rise in memory and storage component costs, Amazon has raised prices on its Echo, Kindle, Fire TV, and Eero devices by up to 60 percent, the company confirmed to Fortune. The hikes affect products across the lineup, with the entry-level Echo Dot jumping from $49.99 to $79.99—a 60 percent increase—and the Kindle base model rising nearly 37 percent to $149.99. The move is the most visible consumer-facing symptom of a broader DRAM shortage that is now driving up costs for everything from AI servers to budget laptops.
The LLVM 23 release brings substantial compile-time improvements, with developers reporting up to 30% faster build times for large codebases, according to community benchmarks and release notes published this week.
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has ordered federal agencies to patch two actively exploited vulnerabilities in TrueConf, a Russian-built video conferencing platform, after compromised servers were found distributing malware to meeting participants. The development comes alongside separate supply-chain attacks targeting Android car head units and a confirmed social engineering breach at cybersecurity firm ReliaQuest, highlighting a surge in multi-vector threats.
A joint advisory from U.S. federal agencies has warned that Iranian hackers are actively targeting Siemens S7-series programmable logic controllers (PLCs) used in critical infrastructure, employing AI tools to develop exploits and adapt to defensive measures. The warning, issued by CISA, NSA, FBI, DOE, and EPA, comes amid ongoing Russian cyber-espionage campaigns against academia, government, and defense organizations, as detailed by Google's Threat Intelligence Group.
In a span of 24 hours, three major organizations — France's tax authority, Toronto's Hospital for Sick Children, and private equity firm Apollo Global Management — disclosed data breaches affecting hundreds of thousands of individuals, underscoring a relentless wave of cyberattacks targeting government agencies, healthcare institutions, and financial giants.
Apple has laid off more than 200 employees from its Vision Pro and Siri teams, according to a Bloomberg report, including the near-shutdown of the Vision Pro gaming team and significant reductions to immersive content production. The cuts come as the company says it is evolving its business to deliver better user experiences, though they raise questions about the future of two high-profile initiatives that have faced mixed reception since launch.
A security researcher who inadvertently hijacked a portion of the global phone number routing system has logged hundreds of thousands of misdirected calls, including many to US military bases, revealing a little-known vulnerability in how telephone numbers are resolved on the internet.
Security researchers at Adversa AI have uncovered a novel prompt injection attack against xAI's Grok chatbot that uses encrypted malicious instructions to evade content filters, enabling the exfiltration of user chat history and personal information. Despite being notified of the vulnerability in June 2026, Grok remains exploitable as of August 2026, raising concerns about the fundamental security of large language models (LLMs).
Microsoft has fixed a maximum-severity vulnerability in its Entra ID identity and access management service that attackers were already exploiting in the wild. The flaw, tracked as CVE-2026-69836 and carrying a CVSS score of 10.0, stems from unsafe deserialization and allows unauthenticated remote code execution over the network. Microsoft disclosed the issue on Thursday, stating that it has been fully mitigated server-side without requiring any action from customers.