01
Security
IT teams scramble to patch critical PHP flaw as no large breaches are confirmed
Patching is slow for sites on older PHP versions, and an early warning of a working exploit was withdrawn.
Patching is slow for sites on older PHP versions, and an early warning of a working exploit was withdrawn.
The system has only been run on historical records, and Microsoft has no commercial timeline for it.
Because the intruders signed in as authorised users, the attack left little for defenders to spot before email and files were exposed.
Enterprise customers gain a second major cloud route to OpenAI's frontier models, while OpenAI keeps Microsoft as its main cloud partner.
Software that depends on web PKI but cannot filter intermediates stays exposed until its updates arrive.
Fake install pages for Claude Code, Codex and NotebookLM show genuine-looking commands while copying malicious ones, a trick that exploits trust in the AI brands themselves.
The biggest change is that agents can now run entirely on a personal machine, keeping user data on the device.
Users can check uploaded images, audio and video at SynthID.com, though a negative result does not prove content is human-made.
Affected users should install the October 2026 security update, which includes the fix for crashes caused by the September preview update.
The update widens access to AMD’s upscaling tech on PC and Xbox, while Xbox still has no launch date.
Admins will be able to fix severe boot failures through Windows Update, with cloud fixes needing to be enabled first on Server 2025.
The 2.25 GW return adds new power for AI-linked data demand, with one plant already moving toward 2027 output.
The pressure now reaches peers like Oracle, as investors reassess how far AI firms can keep borrowing.
Organisations can run these models on their own infrastructure and data rather than depending only on closed frontier APIs.
Microsoft has tied the problem to Windows Server security patches, not the Exchange update itself, and has issued a registry workaround while it fixes the issue.
The update is production-ready for testing and shifts work across runtime, ASP.NET Core, SDK, NuGet, MAUI and Windows Forms without adding new C# features.
The arrangement puts an outside evaluator inside Anthropic’s office as pressure grows for stronger independent oversight of advanced AI.
Buyers can now order RTX Spark laptops, with Best Buy units due on October 16 and some direct models reaching 128GB of unified memory.
UTC will drift away from Earth’s rotation for now, reducing the risk of future timing-related outages and avoiding a leap second adjustment path the BIPM says is too fragile.
The rebuilt apps are meant to make Arm-based Windows laptops faster for creative work and more responsive in daily editing.
Tom’s Hardware says the Windows on Arm laptop is the fastest system it has tested for local agentic AI workloads.
Strong AI-related orders are keeping ASML booked far ahead, while higher prices are set to reach customers from 2027.
The method is meant to make medium-range forecasts easier to refine without large GPU systems.
The takedown cuts off a large source of scam traffic that had reached Microsoft customers and other organizations across more than 150 countries.
The updates also left some devices unable to turn on Smart App Control, while Microsoft says there are temporary workarounds for the other issues.
The update also closes three zero-days and marks the last security update for Windows 11 24H2 Home and Pro and Windows 10 Enterprise LTSB 2016.
Synacor patched the bug before disclosing it, and exposed Zimbra servers are now being scanned at scale.
Datatilsynet says the penalty is its largest so far and gives Xandr three weeks to appeal.
The government is turning to advanced AI to harden systems after saying a breach should be assumed at some point.
Microsoft’s CEO wants AI systems built with stronger controls, records and a human stop button as concern grows over losing control of models.
Older Windows machines will stop getting updates after certificate rotation unless they are moved to a supported release.
The move cuts off new and pending PERM applications for named companies, disrupting a key step H-1B workers use to seek permanent status.
The London trial extends Uber’s multi-partner autonomous vehicle push in Europe and could feed into wider deployment plans.
Teams meetings will be able to surface signals from certified providers, giving organizations warnings when audio, video or identities look suspicious.
Microsoft’s new AI-focused PCs and Windows 11 features point more work toward local, on-device use.
The change makes common settings and actions faster to reach from the taskbar, with Copilot gaining broader access in the months ahead.
The change will stop .msix and .msixbundle files from being sent, received, opened or downloaded by default in Outlook on the web and new Outlook for Windows.
The new site expands provenance checks beyond Google-generated content and adds partner marks from OpenAI, NVIDIA, Kakao and soon Apple.
KB5124010 can break older software that uses AC-3 audio decoding, and Microsoft has not given a fix date yet.
The release keeps more run data in the dashboard and widens the platform’s language and cloud support.
The fix closes a storage-layer leak in Workers Paid containers and shows that isolation still depends on what happens beneath microVM boundaries.
The approach could make new defensive lasers easier for sailors to learn and cheaper to use than missile intercepts.
The default change should cut browser sign-in risk, but companies that rely on code-injecting tools may need to adjust before the rollout ends in late October.
The new delivery chain reduces victim interaction and has already been used in at least 13 phishing campaigns against more than 100 organisations.
Microsoft says the new setup cuts the setup step for isolated containers and is now available in more than 40 Azure regions.
Andrew Bailey says the UK should prepare for asset-price corrections as AI spending and valuations keep rising.
Microsoft says the posts were unauthorized and the account is secured while it investigates the takeover.
The release shifts waypoint support in ambient mesh and forces teams using Istio’s old Google-hosted images and charts to move before the 13 October outage test.
Open source maintainers are now seeing probing begin minutes after a fix is posted, adding pressure to disclosures already moving faster than many systems can patch.
A consolidated Copilot and new Autopilot agent push Microsoft’s AI efforts deeper into everyday work tools.
Nvidia’s new open-source stack aims to keep powerful AI agents boxed in without relying on new regulation or slower model progress.
The campaign extends from tenant-wide resource wipes to direct hits on AI datasets and vector stores.
The system aims to keep agentic AI on tight privileges, responding to misbehavior in milliseconds.
The change is intended to speed responses, but employees will review proposed removals before records are released.
A forged token let the researcher reach internal employee and Bing analytics data before Microsoft fixed the flaw via its bug bounty program.
Teams can discover compatible software in one directory, while partners gain a route to Claude customers.
Affected users may need to reactivate their purchased Office product, but the suspected installation issue remains unconfirmed.
Affected virtual-desktop users may need a manual step to regain access, while enterprise customers have a mitigation.
The newly launched platform lets AI agents act as full participants in company chats, already drawing early users and funding.
The funding backs Ema’s push to replace parts of traditional SaaS and IT services with task-based AI automation.
The Norway hosting setup exposes how Chinese-backed firms can lawfully tap top Nvidia chips abroad despite U.S. export limits.
Vibe coding is shifting from prototyping tool to core enterprise infrastructure as Lovable’s usage and funding surge.
Sensor-signed captures and Apple-only signatures could reshape how image authenticity is verified and who controls that process.
The takedown hits a service that automated device-code phishing at scale, largely against enterprise Microsoft accounts worldwide.
The attack turns trusted AI assistants into high‑privilege proxies once a rogue Chrome‑style extension is installed.
Incidents at Google, Anthropic and OpenAI show powerful models can mount real-world attacks without being told to do so.
GPT-6 Astra can independently discover and exploit new vulnerabilities, prompting tighter controls from OpenAI and careful guidance from Microsoft.