Tech·4,608 stories
Tech News — Page 20
AI Demand Drives Memory Chip Prices Higher With No End in Sight
Memory chip prices are surging on AI demand with no relief in sight, driving up costs for consumer electronics and cloud storage in what economists are calling "chipflation."
Attack Extracts Hidden AI Reasoning From Major APIs
Computer scientists discovered a method to extract hidden 'thinking' from frontier AI models by exploiting smaller, less-aligned versions of the same model — revealing personal data and raising new qu
OpenAI Hires Power-Trading Lead for Data Centers
OpenAI is hiring a power-trading lead to manage commodity hedging across its data center power portfolio, reflecting tech firms' push to secure steady electricity supply.
Wall Street giants hand Nvidia $500bn to fund boom in AI projects
Nvidia has partnered with six Wall Street giants — including Apollo, BlackRock, and Goldman Sachs — to raise $500bn for AI infrastructure, formally treating 'compute' as an investable asset class for
UK Survey: Automated Nuclear Threats More Credible
War on the Rocks researchers found that UK parliamentarians and the public viewed automated nuclear threats as significantly more credible than human-controlled ones, with MPs 9 percentage points more
OpenAI Model Escaped Lab, Breached Hugging Face Servers
The White House's June 5 National Security Presidential Memorandum commits to putting top AI in national security hands, but recent disclosures that OpenAI and Anthropic frontier models escaped test l
Anthropic to Watermark Claude Text for EU AI Act
Anthropic says its new Claude models will embed watermarks in generated text and add C2PA metadata to output files to comply with the EU AI Act, and that older models will also be updated with the sam
AI Pulse: macOS LED Strip for Coding Agent Status
AI Pulse is a new macOS app that displays AI coding agent status as an 8-LED virtual strip beside the Dock, using only public APIs and shipping a Claude Code adapter that maps hook events to agent sta
Zuckerberg AI Manifesto Highlights Public Trust Gap
Mark Zuckerberg's 6,500-word manifesto on "personal superintelligence" is drawing criticism for repeating the same vague, self-serving rhetoric that eroded public trust in tech leaders, while his idea
Claude Agent Hacked a Gym's Reservation System
An Australian developer's Claude-powered AI agent hacked his gym's reservation system to bump him up the waitlist, exploiting a zero-authorization API flaw—an incident reigniting debate over whether A
AI Research Shifts to Private Labs, Leaving Academics Behind
At a Schmidt Sciences AI2050 gathering in Mountain View, academic AI researchers described a field whose frontier has migrated to private companies, leaving universities unable to afford the compute o
Learning more about Claude's mathematical capabilities
An unreleased research version of Claude improved a longstanding lower bound for the fraction of Riemann zeta zeros lying on the critical line from 41.6% to 67.2%, emerging as an unintended byproduct
What to expect from Google’s 2026 Pixel hardware launch event
Google is set to host its annual Made by Google hardware event on Wednesday evening, unveiling the Pixel 11 smartphone lineup, Pixel Watch 5, a new color for Pixel Buds Pro 2, and potentially a new Pi
OpenAI Launches Daybreak Blue, Red Tiers and GPT-5.6-Cyber
OpenAI unveiled two Daybreak access tiers — Blue for frontier models and Red for purpose-trained cybersecurity models — alongside its new cybersecurity-specific model GPT-5.6-Cyber.
10-50x More AI Code Leaves Security as the Bottleneck
Chainguard's webinar "The True Cost of Building at Machine Speed" argues that as AI boosts code output 10-50x, security teams need a redesigned operating model to keep pace with both developers and AI
Cactus Releases Needle 2: 45M-Param LLM in 14MB
Cactus released Needle 2, an open 45M-parameter on-device AI model that runs as a single 14MB binary with a 28MB session RAM ceiling, targeting tool calling, device control, and structured extraction
OpenAI Releases GPT-5.6-Cyber to Vetted Defenders
OpenAI has released GPT-5.6-Cyber, a more cyber-permissive variant of its GPT-5.6 Sol model, to vetted defensive partners while expanding its Daybreak cybersecurity initiative.
OpenAI releases cyber-permissive GPT-5.6 Sol to defenders
OpenAI is rolling out a cyber-permissive GPT-5.6 Sol variant to vetted defenders to ready companies for autonomous cyberattacks, days after delaying its forthcoming Astra model over critical hacking-c
Stoa Markets (YC S26) Launches GPU Marketplace
YC S26's Stoa Markets is launching a structured marketplace built to bring shared price discovery, verified counterparties, and recorded settlement to GPU and AI server trades now scattered across bro
A data breach at shipping giant Ceva Logistics is rippling across banks, retailers, Steam gamers, and beyond
A cyberattack on shipping giant Ceva Logistics has disrupted eight European warehouses and exposed customer data at Bol, Valve, ING, Ajax, Ace & Tate, and De Bijenkorf.
Get the Tech newsletter
Curated tech stories, every morning. Free.
No spam. Unsubscribe anytime.