<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
<title>Park Labs — AI news and lab notes</title>
<link>https://park-labs.com/en/</link>
<atom:link href="https://park-labs.com/en/feed.xml" rel="self" type="application/rss+xml"/>
<description>A solo AI and automation lab. Beginner-friendly AI news and open lab notes on building, revenue and failures.</description>
<language>en</language>
<lastBuildDate>Sat, 07 Nov 2026 00:00:00 GMT</lastBuildDate>
<item><title>OpenAI introduces GPT‑6 Cyber for approved security testing</title><link>https://park-labs.com/en/news/daily/2026-11-07-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-07-no-01/</guid><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate><description>Two vendor tests measure different things. The release also separates initial access from fixes on a reader’s own device.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Security</category></item>
<item><title>Google extends Gemini Omni 1.2 Flash videos to 15 seconds</title><link>https://park-labs.com/en/news/daily/2026-11-07-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-07-no-02/</guid><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate><description>A fictional shop shows what fits in five extra seconds, how app rollout differs from API access and how to estimate 720p output costs.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Google</category><category>Video Generation</category></item>
<item><title>OpenAI reports 948 proposed Linux fixes: the evidence people must review</title><link>https://park-labs.com/en/news/daily/2026-11-06-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-06-no-01/</guid><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Security</category><category>Research</category></item>
<item><title>OpenAI announces Images 2.6: preserve details, target edits and read GIF and cost conditions</title><link>https://park-labs.com/en/news/daily/2026-11-06-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-06-no-02/</guid><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Image Generation</category></item>
<item><title>CISA lists exploitation of an Apple WebKit flaw: check the existing fixes</title><link>https://park-labs.com/en/news/daily/2026-11-05-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-05-no-01/</guid><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Apple</category><category>Security</category></item>
<item><title>Node.js plans November 11 security releases: prepare the running environments</title><link>https://park-labs.com/en/news/daily/2026-11-05-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-05-no-02/</guid><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Node.js</category><category>Security</category></item>
<item><title>OpenAI plans to acquire code-review toolmaker Pierre: existing tools and future integration</title><link>https://park-labs.com/en/news/daily/2026-11-05-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-05-no-03/</guid><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Open Source</category></item>
<item><title>Anthropic details text watermarking for three Claude models: what detection can establish</title><link>https://park-labs.com/en/news/daily/2026-11-05-no-04/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-05-no-04/</guid><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Anthropic</category><category>Provenance</category></item>
<item><title>CISA lists exploitation of a WordPress upgrade-window flaw</title><link>https://park-labs.com/en/news/daily/2026-11-04-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-04-no-01/</guid><pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>WordPress</category></item>
<item><title>IPA extends PHP checks to VPN and security appliances: trace fixes into the device</title><link>https://park-labs.com/en/news/daily/2026-11-04-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-04-no-02/</guid><pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>PHP</category></item>
<item><title>OpenAI reports Astra device design and automotive research: who made the prototype?</title><link>https://park-labs.com/en/news/daily/2026-11-04-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-04-no-03/</guid><pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>OpenAI</category><category>Research</category><category>Engineering</category></item>
<item><title>OpenAI explains Pando: where code from parallel AI work goes</title><link>https://park-labs.com/en/news/daily/2026-11-04-no-04/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-04-no-04/</guid><pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>OpenAI</category><category>Engineering</category></item>
<item><title>OpenAI introduces LifeVal: reading task scores and evidence of completion</title><link>https://park-labs.com/en/news/daily/2026-11-03-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-03-no-01/</guid><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>OpenAI</category><category>Research</category><category>Evaluation</category></item>
<item><title>OpenAI behavior guidance explained: can a webpage authorize an AI action?</title><link>https://park-labs.com/en/news/daily/2026-11-03-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-03-no-02/</guid><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>OpenAI</category><category>Agents</category><category>Security</category></item>
<item><title>Karpathy explains HackerTube: the AI work and the resources people prepared</title><link>https://park-labs.com/en/news/daily/2026-11-02-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-02-no-01/</guid><pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Karpathy</category><category>Claude</category><category>Agents</category><category>Web</category></item>
<item><title>Applied fixes, AI work and learning: how to check a delegated result</title><link>https://park-labs.com/en/news/weekly/2026-11-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/weekly/2026-11-02/</guid><pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate><description>PHP package rebuilds and GMO customer contact leave concrete follow-up checks. Computer use, simulated scenes, mathematics tutoring and video tools also need different evidence of their results.</description><category>weekly</category><category>AI News</category><category>Weekly</category><category>Security</category><category>Open Source</category></item>
<item><title>OpenAI backs a common training-preference signal; search and access stay separate</title><link>https://park-labs.com/en/news/daily/2026-11-01-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-11-01-no-01/</guid><pubDate>Sun, 01 Nov 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>OpenAI</category><category>Web</category><category>Policy</category></item>
<item><title>AI work, server fixes and customer notices: October’s changes and evidence</title><link>https://park-labs.com/en/news/monthly/2026-11-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/monthly/2026-11-01/</guid><pubDate>Sun, 01 Nov 2026 00:00:00 GMT</pubDate><description>October coverage connects AI computer work, PHP/VPN fixes, GMO customer guidance and a security-audit skill. The synthesis explains the evidence needed for permissions, application, deposits and investigations after announcements.</description><category>monthly</category><category>AI News</category><category>Monthly</category><category>Security</category><category>Open Source</category></item>
<item><title>Ivanti VPN chain adds exploited flaws; recheck October 13 patches</title><link>https://park-labs.com/en/news/daily/2026-10-31-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-31-no-01/</guid><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>Security</category><category>Ivanti</category><category>VPN</category></item>
<item><title>GMO starts customer contact on payments and replacement desktops; completion still needs evidence</title><link>https://park-labs.com/en/news/daily/2026-10-31-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-31-no-02/</guid><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>GMO</category><category>Cloud</category></item>
<item><title>OpenAI details agent monitoring: an alert does not always mean automatic stopping</title><link>https://park-labs.com/en/news/daily/2026-10-31-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-31-no-03/</guid><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>OpenAI</category><category>Research</category></item>
<item><title>CISA adds exploited GitLab regexp flaw: verify the running fix and intrusion evidence</title><link>https://park-labs.com/en/news/daily/2026-10-30-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-30-no-01/</guid><pubDate>Fri, 30 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>GitLab</category></item>
<item><title>Exploited Cisco VPN flaw can interrupt work before login succeeds</title><link>https://park-labs.com/en/news/daily/2026-10-30-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-30-no-02/</guid><pubDate>Fri, 30 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>Cisco</category></item>
<item><title>GMO reports potentially identifying leads as submissions reach 321</title><link>https://park-labs.com/en/news/daily/2026-10-30-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-30-no-03/</guid><pubDate>Fri, 30 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>GMO</category><category>Cloud</category></item>
<item><title>Google and Eedi study 1,179 pupils: tutoring gains differ across tasks and broad tests</title><link>https://park-labs.com/en/news/daily/2026-10-30-no-04/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-30-no-04/</guid><pubDate>Fri, 30 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Google</category><category>Education</category><category>Research</category></item>
<item><title>CISA flags Linux flaw that can compound PHP attacks into root access</title><link>https://park-labs.com/en/news/daily/2026-10-29-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-29-no-01/</guid><pubDate>Thu, 29 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>PHP</category><category>Linux</category></item>
<item><title>GMO confirms publication-site domain suspension and updates device-reuse guidance</title><link>https://park-labs.com/en/news/daily/2026-10-29-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-29-no-02/</guid><pubDate>Thu, 29 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>GMO</category><category>Cloud</category></item>
<item><title>Anthropic and Enveda detail AI-assisted checks of drug-manufacturing records</title><link>https://park-labs.com/en/news/daily/2026-10-29-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-29-no-03/</guid><pubDate>Thu, 29 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Anthropic</category><category>Science</category></item>
<item><title>GMO cloud breach: purported customer data published as device investigation continues</title><link>https://park-labs.com/en/news/daily/2026-10-28-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-28-no-01/</guid><pubDate>Wed, 28 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>Security</category><category>GMO</category><category>Cloud</category></item>
<item><title>PHP flaws enter CISA’s exploited-vulnerability list: check when fixes took effect</title><link>https://park-labs.com/en/news/daily/2026-10-28-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-28-no-02/</guid><pubDate>Wed, 28 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Security</category><category>PHP</category><category>WordPress</category></item>
<item><title>Google previews Genie 4, generating scenes and sound that respond to user control</title><link>https://park-labs.com/en/news/daily/2026-10-28-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-28-no-03/</guid><pubDate>Wed, 28 Oct 2026 00:00:00 GMT</pubDate><description>Google DeepMind announced Genie 4 on October 27. The world model generates scenes, sound and dialogue in response to user control. Access is limited to trusted testers, with subscriber and developer availability still planned.</description><category>daily</category><category>AI News</category><category>Google</category><category>Genie</category></item>
<item><title>PHP rebuilds Windows packages with a Brotli security fix</title><link>https://park-labs.com/en/news/daily/2026-10-27-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-27-no-01/</guid><pubDate>Tue, 27 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>Security</category><category>PHP</category><category>Brotli</category></item>
<item><title>GPT-6.1 Terra launches: the cost and checks behind computer work</title><link>https://park-labs.com/en/news/daily/2026-10-27-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-27-no-02/</guid><pubDate>Tue, 27 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>OpenAI</category><category>GPT-6.1 Terra</category></item>
<item><title>Anthropic publicly files S-1 for proposed IPO; price and share count unset</title><link>https://park-labs.com/en/news/daily/2026-10-27-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-27-no-03/</guid><pubDate>Tue, 27 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Anthropic</category><category>IPO</category></item>
<item><title>How Cloudflare’s skill verifies an AI security warning</title><link>https://park-labs.com/en/news/daily/2026-10-26-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-26-no-01/</guid><pubDate>Mon, 26 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Cloudflare</category><category>Security</category><category>Open Source</category></item>
<item><title>Connecting AI to work, then verifying what runs</title><link>https://park-labs.com/en/news/weekly/2026-10-26/</link><guid isPermaLink="true">https://park-labs.com/en/news/weekly/2026-10-26/</guid><pubDate>Mon, 26 Oct 2026 00:00:00 GMT</pubDate><description>Server operators need evidence of applied fixes; teams introducing AI need to check connections, sources and permissions. This synthesis of 14 dailies published October 19–25 also shows how a 10× faster judgment changes the total wait.</description><category>weekly</category><category>AI News</category><category>Weekly</category><category>Security</category><category>Open Source</category></item>
<item><title>Checking business information with ChatGPT, from reference to source</title><link>https://park-labs.com/en/news/daily/2026-10-25-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-25-no-01/</guid><pubDate>Sun, 25 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>ChatGPT</category><category>Small Business</category></item>
<item><title>PHP adds upload-progress checks after its patches</title><link>https://park-labs.com/en/news/daily/2026-10-24-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-24-no-01/</guid><pubDate>Sat, 24 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Security</category><category>PHP</category><category>WordPress</category></item>
<item><title>OpenAI details stronger isolation after research AI reached the web</title><link>https://park-labs.com/en/news/daily/2026-10-24-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-24-no-02/</guid><pubDate>Sat, 24 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Security</category><category>AI Agents</category></item>
<item><title>OpenAI explains a 120ms-to-10ms storage-read improvement</title><link>https://park-labs.com/en/news/daily/2026-10-24-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-24-no-03/</guid><pubDate>Sat, 24 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Infrastructure</category></item>
<item><title>OpenAI makes Decisions API generally available for fast, bounded judgments</title><link>https://park-labs.com/en/news/daily/2026-10-23-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-23-no-01/</guid><pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>API</category></item>
<item><title>Anthropic donates the Agent Skills standard to a neutral foundation</title><link>https://park-labs.com/en/news/daily/2026-10-23-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-23-no-02/</guid><pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Anthropic</category><category>Agent Skills</category><category>Open Source</category></item>
<item><title>OpenAI explains FDE the work between an AI demo and daily operations</title><link>https://park-labs.com/en/news/daily/2026-10-23-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-23-no-03/</guid><pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>FDE</category><category>Codex</category></item>
<item><title>Node.js ships security releases fixing a remote code execution flaw</title><link>https://park-labs.com/en/news/daily/2026-10-22-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-22-no-02/</guid><pubDate>Thu, 22 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Security</category><category>Node.js</category><category>Open Source</category></item>
<item><title>OpenAI models enter private preview for selected Google Cloud enterprises</title><link>https://park-labs.com/en/news/daily/2026-10-22-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-22-no-03/</guid><pubDate>Thu, 22 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>OpenAI</category><category>Google</category><category>Google Cloud</category></item>
<item><title>PHP announces critical patches for two vulnerabilities</title><link>https://park-labs.com/en/news/daily/2026-10-20-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-20-no-01/</guid><pubDate>Wed, 21 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Security</category><category>PHP</category><category>WordPress</category></item>
<item><title>Hugging Face releases a library for connecting multiple AI models</title><link>https://park-labs.com/en/news/daily/2026-10-21-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-21-no-01/</guid><pubDate>Wed, 21 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>AnyLanguageModel</category><category>Open Source</category></item>
<item><title>Google announces Gemini tools for small-business visuals and finances</title><link>https://park-labs.com/en/news/daily/2026-10-21-no-02/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-21-no-02/</guid><pubDate>Wed, 21 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Google Gemini</category></item>
<item><title>ChatGPT adds locked chats to hide saved conversations from view</title><link>https://park-labs.com/en/news/daily/2026-10-21-no-03/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-21-no-03/</guid><pubDate>Wed, 21 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>ChatGPT</category></item>
<item><title>PHP releases emergency patches for two critical vulnerabilities</title><link>https://park-labs.com/en/news/daily/2026-10-22-no-01/</link><guid isPermaLink="true">https://park-labs.com/en/news/daily/2026-10-22-no-01/</guid><pubDate>Wed, 21 Oct 2026 00:00:00 GMT</pubDate><description>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.</description><category>daily</category><category>AI News</category><category>Daily</category><category>Security</category><category>PHP</category><category>WordPress</category><category>Open Source</category></item>
</channel>
</rss>
