- Security
PHP security fixes are out. Check which build serves your site
PHP released fixes on November 19. Find the release for your running branch and confirm the package and application time at your site.
Read the article
Loading Park Labs...
PHP released fixes on November 19. Find the release for your running branch and confirm the package and application time at your site.
Read the article
An authenticated report can reach an unsafe processing path. Check the fixed branch, restart, and version actually running on each server.
Drupal released 11.4.13, 11.3.22, and 10.6.22. Follow the installer-protection guidance and obtain evidence of application on the live site.
PHP has announced security releases for all supported versions. Detailed conditions and fixed version numbers will accompany the release.
Compare app access with API use, see who changes a spreadsheet, and follow a fictional bill that holds token volumes constant.
Drupal’s security team has announced a forthcoming security update for supported branches. Check which version your site runs and who can update it.
Alibaba’s headline price applies to reused input in Singapore. A worked example shows how new input and output affect the bill.
The partnership aims to support up to 25,000 young people. Here is what the announcement says about training and what applicants still need to check.
Check the running version, the new full backup and the plan for moving off PostgreSQL 14.
Check Point observed attacks against remote-access gateways. Here is what proves a fix is running and what helps establish whether a device was already compromised.
The chat server’s maintainers reported exploitation. One network name can cover several servers, so operators need evidence from each one.
OpenAI is rolling out agent-assisted payment to some US Pro accounts. Saving a card, approving one purchase and getting a confirmed order are separate steps.
Apache has fixed a critical flaw in its HTTP/2 component. Here is why the upstream and operating-system packages can carry different numbers and what shows that the fix is running.
The free Gemini app may choose Hydrogen automatically for a prompt. Developers can name Hydrogen directly in free or paid Gemini API projects.
Node.js released security fixes for its 22, 24 and 26 branches. A deployment example shows how to check whether the servers handling requests are running the fix.
Chrome’s desktop fix is rolling out. Here’s how to check the running version and see whether you need to relaunch.
Google opened Carbon to paid API users and began rolling it out in the Gemini app for AI Ultra subscribers. One document task shows how access and December pricing work.
OpenAI plans to remove GPT Image 2 from the API on February 10, 2027. Here is how to find out whether an app uses it and compare the two replacements on the same task.
Claude Fable 5.2 and Mythos 5.2 have different access rules. Compare the Claude app’s plan conditions and the 20% lower standard API rates on the same task.
Red Hat released OpenShift 4.19.55 on November 9. Here is how to check whether the fix for the exploited Brotli flaw has reached your cluster and apps.
A false size label in an icon can corrupt an image server’s working memory. Separate the earlier fixes, the November 8 CVE records and actual deployment.
Valid credentials for another sign-in method can bypass the company’s sign-in checks and may bypass extra verification. The fixed build depends on the firewall generation.
Two vendor tests measure different things. The release also separates initial access from fixes on a reader’s own device.
A fictional shop shows what fits in five extra seconds, how app rollout differs from API access and how to estimate 720p output costs.
OpenAI’s November 4 Linux study reports 948 proposed privilege-escalation patches after controlled validation. Maintainer review, supplied packages and running-system application are separate stages.
OpenAI announced Images 2.6 on November 4 with Flare, Sunburst and more editing controls. GIF creation is rolling out in ChatGPT; scene complexity and rendering quality are distinct API settings.
CISA added an Apple WebKit flaw to its exploitation catalog on November 4. The read/write prerequisite, historical 2024 fixes and current device records explain what an operator can check.
Node.js plans security releases for 26.x, 24.x and 22.x on November 11 or shortly after. The advance notice gives operators time to map environments, support status and evidence of later application.
OpenAI announced its planned Pierre acquisition on November 4. Trees and Diffs already used in Codex, separate code storage and future integration have different roles and delivery states.
Anthropic’s November 3 notes describe text watermarking for three Claude models. Text marks, file credentials and delivery routes offer clues to processing, with limits on authorship and accuracy judgments.
CISA added a WordPress-core upgrade code-execution flaw to KEV on November 3. The October 28 fix, upgrade records and investigation of earlier activity need separate evidence.
IPA added appliance web interfaces to its PHP warning on November 2. Product exposure, vendor fixes and evidence that an individual device is running the remedy require separate records.
OpenAI described Astra-assisted device design and manufacturing research on November 3. The design, manufacturing review, prototype and BenchCAD results mark different stages and evidence.
In a November 2 technical article, OpenAI introduced its internal code-delivery system Pando. Its parallel transfers, human and AI roles, and nearly 80% reduction in push latency need distinct interpretations.
OpenAI introduced its LifeVal evaluation of everyday AI tasks on November 2. Its average scores, best recorded task results and a kindergarten shortlist need different evidence of what was actually completed.
OpenAI’s October 29 behavior guidance treats outside instructions as information by default. A fictional hotel comparison shows who can authorize booking and which execution records matter.
This November 2 follow-up checks Karpathy’s October 31 HackerTube post and added process account. It explains selection criteria, human-prepared access/server/domain resources and the AI’s development work.
OpenAI said on October 30 it will honor a common signal for training preferences, targeting implementation in coming months. Search controls and actual protection of private material remain separate.
CISA added two Ivanti Connect Secure flaws to its exploited-vulnerability catalog on October 30. Match the vendor’s October 29 fixes to actual appliances; earlier credential theft and data loss need separate investigation.
GMO Internet says it began contacting customers on October 30 about goodwill payments and replacement desktops. The ¥5000 payment, trading-loss compensation and consent to migrate have separate conditions.
OpenAI described its system for detecting unauthorized agent actions on October 29. In the API, automatic stopping depends on conversation tracking even when requests are monitored. Alerts, stopping and undoing work are separate states.
CISA added GitLab’s unauthenticated remote code execution flaw to its exploited-vulnerability catalog on October 29. The patch dates to October 12. Self-managed operators need both running-version evidence and a review for intrusion traces.
CISA added the Cisco ASA/FTD VPN reload flaw to its exploited catalog on October 29. Exposure depends on affected software, reachable SSL remote-access VPN and secondary-authentication settings together. Verify the fix on the running device.
GMO Internet said on October 29 that its own investigation obtained information it believes could help identify the perpetrators and would submit it to police. Submissions reached 321 at 18:00. Obtaining leads, intended submission and investigative findings are separate stages.
Google and Eedi randomly assigned 1,179 UK pupils to four support conditions. Experts reviewed every AI message before delivery. Gains on nearby learning tasks did not become an overall advantage on the broad test after ten weeks.
CISA added Linux kernel flaw CVE-2026-46300 to its exploited-vulnerability catalog on October 28, describing a possible chain with existing PHP flaws. Operators need separate evidence for web PHP and the running server kernel.
GMO Internet reported a suspended publication-site domain on October 28 and 113 submissions to its information channel. Notice 14 says no infection traces were found in examined devices and asks customers to scan and back up before reuse.
Anthropic published early drug-development results on October 27. Enveda’s document-review case puts human source checks after AI comments. Its measured time savings concern a defined review task, while treatment efficacy requires separate evidence.
Customers who accessed GMO Internet’s Onamae.com Desktop Cloud should use another safe device to protect trading passwords, end existing connections and check transactions. The company confirms data removal; individual device infection and the technical entry path remain under investigation.
CISA added two existing PHP flaws to its exploited-vulnerability catalog on October 26; Japan’s IPA updated its notice on October 27. This changes the threat picture, not the patch version. Check web PHP application and the earlier exposure period.
Google DeepMind announced Genie 4 on October 27. The world model generates scenes, sound and dialogue in response to user control. Initial access was limited to trusted testers, with subscriber and developer availability still planned.
The PHP team announced rebuilt Windows packages on October 26 with Brotli 1.2.1. The runtime number is unchanged, so operators need the package revision and evidence that the web process uses it.
OpenAI introduced GPT-6.1 Terra on October 26 for visual reasoning, computer use and writing. Standard API token rates are half those of GPT-6.1 Sol; permissions and result checks still shape an actual workflow.
Anthropic publicly filed an IPO registration statement on October 26. Following its June confidential draft, readers can now examine its business and risk disclosures; offering price and share count remain unset.
Cloudflare’s public security-review skill is a work manual for coding AI. A fictional invoice service shows what evidence is needed to check an AI’s suspicion before deciding what to fix.
A practical follow-up to OpenAI Forum’s small-business demonstrations: compare menus and event listings against a reference, retain source evidence and leave final corrections to a person.
PHP issued additional guidance on October 23 for a specific combination of session settings and code. Server operators should ask their host about progress tracking and runtime format changes, and keep existing tracking disablement until the next regular fix.
On October 23, OpenAI described how it hardened training environments after an internal research AI gained limited external access. Its approach combines saved web content for practice with permissions, execution isolation, network controls and monitoring.
On October 23, OpenAI explained performance work on its Habitat storage platform. It reports median storage-read latency falling from roughly 120ms to 10ms; that is not a 12-fold speedup of entire ChatGPT responses.
OpenAI announced general availability of the Decisions API on October 22. Developers can get a predefined choice, probability or score from text, images and JSON. Its speed claim compares the same GPT-6 Luna through the Responses API.
Anthropic announced on October 22 that it is donating Agent Skills to the Linux Foundation’s Agentic AI Foundation. The format packages procedures and resources for recurring work. Moving between tools still requires checking integrations, permissions and execution environments.
OpenAI's October 22 article explains how its FDE team works alongside customers, tests whether AI can do a real job, and makes successful solutions reusable. A fictional returns desk shows the connections needed between a convincing reply and a completed task.
Node.js released versions 22.23.4, 24.22.1 and 26.12.1 on October 21 to fix security flaws. Crafted Brotli data can lead to code execution; damage depends on the affected program’s permissions.
The PHP team released 8.5.12, 8.4.27, 8.3.36 and 8.2.35 on October 21, fixing two remote code execution vulnerabilities. End-of-life 7.4, 8.0 and 8.1 got only minimal source backports, and WordPress operators need to ask their host whether the fix has reached the server's web-facing PHP.
OpenAI and Google Cloud announced on October 21 and October 22 respectively that OpenAI models are coming to Google Cloud's Model Garden. For now, only selected enterprise customers can use GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna, in private preview.
Hugging Face released AnyLanguageModel 1.0, a library for Swift developers, on October 20. It lets an app send requests to on-device models and hosted AI services the same way, but developers still choose the model and cover any hosted service's fees themselves.
Google announced Gemini tools for small businesses on October 20. They remember a shop's style for marketing material and turn its records into financial charts, rolling out globally to all Gemini app users over the coming days.
ChatGPT is rolling out locked chats. A locked conversation drops out of your regular list and search, and only your PIN or Face ID opens it again. Reset the PIN and the locked chats are deleted for good.
A record from before release, when PHP announced fixes for two vulnerabilities rated 9.2 for October 21 at 23:00 JST/KST. The fixes have since been published; part 2 covers the actual changes. Updating WordPress core or plugins doesn't change the server's PHP, so operators should look up their PHP version under Site Health and ask their host when the fix will be applied.