Breaking news and updates from the world of technology.
A joint advisory from NSA, CISA, FBI, DOE, and EPA describes an active campaign using AI-generated Python scripts to attack Siemens S7 industrial controllers across six critical sectors.
Waymo ends the waitlist for its Ojai purpose-built robotaxi in LA, Phoenix, and San Francisco, days after winning the biggest regulatory approval in its history.
Prime Air is scaling roughly sixfold from its current seven-state footprint, adding 11 new launch sites across Arizona, Florida, Kansas, Louisiana, Michigan, Nebraska, and Texas, with Georgia, Ohio, Illinois, Idaho, and New York coming next.
T-Mobile's cybersecurity chief and three colleagues traveled to a Bellevue data center and snipped a cable by hand to sever a compromised connection during the Salt Typhoon intrusions — avoiding the large-scale breach that hit AT&T, Verizon, and other carriers.
Philadelphia biotech startup Vivodyne has launched a network of robotic HIVE labs that grow living human tissue and generate causal biological data — the missing ingredient AI drug-discovery models need to move past mouse trials.
CVE-2026-33824 lets unauthenticated attackers execute code on any Windows 10, 11, or Server system via a double-free bug in the IKE VPN component — CISA added it to its exploited-vulnerabilities catalog on August 18.
CVE-2026-19478, rated CVSS 9.4, allows unauthenticated remote attackers to destroy public GitLab projects — the third critical GraphQL bug the platform has patched this year.
A promotional video discovered inside macOS Tahoe 26.7 Release Candidate shows Apple's camera-integrated earbuds using Visual Intelligence — the clearest evidence yet of an imminent September launch.
North Korea's Lazarus Group has been actively exploiting a Windows kernel zero-day (CVE-2026-68820) as part of Operation Dream Job, deploying a new rootkit variant and a previously undocumented backdoor against defense and aerospace targets.
Stripe has finalized a deal to buy the AI model routing startup OpenRouter for over $7 billion, gaining control of the most widely used neutral gateway to 400-plus AI models.
What began as a performance incident at 13:40 UTC cascaded through API requests, Actions, Issues, Pull Requests, authentication, and Copilot over more than seven hours, with error rates hitting 50% on repository downloads at the peak.
The cybersecurity giant is buying an AI-agent security startup that stress-tests autonomous systems across 50+ sandboxed environments, extending its FortiAIGate protection from language models to the agents and tools built on top of them.