2026-10-23 notis CATEGORY Join
Show Read as Find Sections Models and releasesAgents and appsResearchMoney and computeLaws and rulingsSafety and securityAI at work More ArchiveModelsAboutCorrectionsThe numbersRSS
My notis
TODAY'S NEWS
October 2026 3 STORIES
MTWTFSS
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Days with a dot have news. See the archive.

The whole week ›
19:27
0/3
WEATHER THEME CARD SPACING

Today's news

today

My page

Kept in this browser only.

SAVED FOLLOWING COLLECTIONS READING SINCE
Nothing saved yet. Tap the bookmark on a card to keep it here. read today
all cases

No case followed yet.

Price changes Stories where a published price went up or down.
Models being retired Models being withdrawn, and gaps before a replacement is ready.
The day's top stories. Sign up
toned down Safety 23 OCT

PHP patches two critical remote code execution flaws

CalifGPTOpenAIPHPWiz
Illustration: A sealed junction box with a hinged service door; its green latch stands for the patch that closes both flaws.Illustration made with AI for ai notis

The PHP project released updates on 21 October for two flaws rated CVSS 9.2, and OpenAI says Calif found a critical flaw using GPT-5.6 Sol.

SOURCEPHP project · 16 sources
CHECKED23 Oct, 09:20
Safety read
Illustration: A sealed junction box with a hinged service door; its green latch stands for the patch that closes both flaws. Safety / 23 OCT

PHP patches two critical remote code execution flaws

  • PHP released 8.5.12, 8.4.27, 8.3.36 and 8.2.35 on 21 October 2026 to fix CVE-2026-109618 and CVE-2026-109619, which can allow unauthenticated remote code execution.
  • PHP rates both flaws CVSS v4.0 9.2. Its advisory says updating is the only complete mitigation, and the mitigations it lists do not make an installation safe.
  • OpenAI says Calif identified a critical flaw using GPT-5.6 Sol, and that Wiz independently discovered and reported both vulnerabilities.

The PHP project published a security advisory and four releases on 21 October 2026: PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35. The releases fix two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, that PHP says can allow unauthenticated remote code execution. PHP rated both CVSS v4.0 9.2. CVE-2026-109618 is described as remote code execution through out-of-memory corruption: PHP uses longjmp() to leave the current context on an out-of-memory error, which can leave request memory invalid for request shutdown to read. CVE-2026-109619 is remote code execution through stale global state, where request-owned internal state can survive after its backing memory is freed and a later request handled by the same worker can use it. PHP says the highest risk is mod_php, with a demonstrated exploit where an unauthenticated request can trigger the attack before any application code runs. PHP lists mitigations but says they do not make an installation safe; updating is the only complete fix. OpenAI says Calif identified a critical flaw using GPT-5.6 Sol as part of OpenAI's Patch the Planet initiative, and that Wiz independently found and reported both vulnerabilities. OpenAI also says it is committing $1 billion in subsidized Daybreak access for organizations with limited security resources.

What it means for you Our view

If you run PHP, the advisory says updating is the only complete fix, so the patched versions are the first thing to check. The highest risk PHP describes is mod_php, where an unauthenticated request can trigger the attack before any application code runs. The mitigations PHP lists do not make an installation safe. The AI-assisted discovery is OpenAI's account: it says Calif used GPT-5.6 Sol and that Wiz found both flaws independently, so PHP's advisory is the better guide to what to patch. Check which PHP version each of your servers and containers runs and compare it with 8.5.12, 8.4.27, 8.3.36 and 8.2.35.

TAGSCalifGPTOpenAIPHPWiz
Compute 23 OCT

OpenAI will bring its models and Codex to Google Cloud

CodexGoogleModel GardenOpenAI
Illustration: A patch panel with rows of ports; its green plug-in module, held just outside an open bay, stands for OpenAI's models that are coming to Google Cloud's Model Garden.Illustration made with AI for ai notis

OpenAI announced the partnership with Google Cloud on 21 October, and says general availability is planned in the coming weeks.

SOURCEOpenAI · 1 source
CHECKED23 Oct, 04:29
Compute read
Illustration: A patch panel with rows of ports; its green plug-in module, held just outside an open bay, stands for OpenAI's models that are coming to Google Cloud's Model Garden. Compute / 23 OCT

OpenAI will bring its models and Codex to Google Cloud

  • OpenAI says it and Google Cloud are bringing OpenAI's frontier models and Codex agentic workflows into Google Cloud's Model Garden, in a post of 21 October 2026.
  • OpenAI says customers will buy the models through the Google Cloud Marketplace, and enterprises with Google Cloud spend commitments will be able to draw those down for the purchases.
  • OpenAI says general availability is planned in the coming weeks and its models will soon be in Google Antigravity and the Gemini Agent. Microsoft remains its primary cloud partner.

OpenAI announced on 21 October 2026 that it is partnering with Google Cloud to bring its frontier models and Codex agentic workflows into Google Cloud's Model Garden. Google Cloud customers will be able to buy the models through Model Garden, transacted through the Google Cloud Marketplace, and enterprises with Google Cloud spend commitments will be able to draw down those commitments for the purchases. OpenAI says the models will soon also be available in Google Antigravity and the Gemini Agent. General availability is planned in the coming weeks, and a request-access form is open. OpenAI states that Microsoft remains its primary cloud partner.

What it means for you Our view

This is OpenAI's announcement, and the only timing it gives is the coming weeks. For a company that buys through Google Cloud, the practical question is the contract: OpenAI says spend commitments can be drawn down against these purchases, but the terms and prices are not part of what we have checked. Until general availability, OpenAI's post points to a request-access form. Ask your Google Cloud account team whether OpenAI models will count toward your commitment, and at what price, before you plan any move.

TAGSCodexGoogleModel GardenOpenAI
OpenAI · 1 source tap to close
Tooling 23 OCT

Anthropic donates Agent Skills to the Agentic AI Foundation

Agent SkillsAgentic AI FoundationAnthropicLinux FoundationSKILL.md
Illustration: An archive cabinet with a drawer of identical folders; its green handover plate stands for ownership of the standard passing to a foundation.Illustration made with AI for ai notis

Anthropic's post of 22 October says the name, logo, GitHub organization and domain move to the foundation, which operates under the Linux Foundation.

SOURCEAnthropic · 1 source
CHECKED23 Oct, 04:29
Tooling read
Illustration: An archive cabinet with a drawer of identical folders; its green handover plate stands for ownership of the standard passing to a foundation. Tooling / 23 OCT

Anthropic donates Agent Skills to the Agentic AI Foundation

  • Anthropic says it is donating the Agent Skills name and logo, the GitHub organization and the agentskills.io domain to the Agentic AI Foundation, which operates under the Linux Foundation.
  • Anthropic says more than 70 agent platforms, including ChatGPT, Cursor, GitHub Copilot and VS Code, have integrated Agent Skills, and nearly 280 million skills have been installed.
  • The proposed roadmap includes versioned releases, a formal proposal process and security guidance for skill authors and installers. Engineers from several companies will maintain the standard as individuals.

Anthropic said on 22 October 2026 that it is donating Agent Skills, an open standard for giving AI agents specialized knowledge and workflows, to the Agentic AI Foundation (AAIF). The donation covers the Agent Skills name and logo, the GitHub organization and the agentskills.io domain. AAIF operates under the Linux Foundation; Anthropic co-founded it last year with Block and OpenAI and has also donated the Model Context Protocol to it. Engineers from Anthropic, Google, OpenAI and other companies will maintain the standard as individuals, and the lead maintainers are David Soria Parra and Mahesh Murag. Anthropic says the proposed roadmap includes versioned releases, a formal proposal process and security guidance for skill authors and installers. Anthropic launched skills in October 2025 and published them as an open standard in December 2025. It says more than 70 agent platforms have integrated the standard, including ChatGPT, Cursor, GitHub Copilot and VS Code, that one million skills have been published, and that nearly 280 million have been installed.

What it means for you Our view

The adoption figures are Anthropic's own. The donation gives the standard a home outside Anthropic: the post says engineers from Anthropic, Google, OpenAI and other companies will maintain it as individuals, with David Soria Parra and Mahesh Murag as Lead Maintainers. For teams that write skills, the part to watch is the proposed security guidance for authors and installers, which is still a proposal. If you install skills from third parties, check where each one comes from and who maintains it, and read its SKILL.md before you enable it.

TAGSAgent SkillsAgentic AI FoundationAnthropicLinux FoundationSKILL.md
Anthropic · 1 source tap to close
NOTHING IN THIS FILTER No story today matches. open the archive →
That's today's news: three stories. ExplainedHow an agent loads a skill in three steps

The morning email

On the mornings we publish: the three top stories and up to four short ones. Free.

We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep

1 of 3
Illustration: A sealed junction box with a hinged service door; its green latch stands for the patch that closes both flaws. Illustration made with AI for ai notis

Safety · 23 OCTnarrowed

PHP patches two critical remote code execution flaws

The PHP project released updates on 21 October for two flaws rated CVSS 9.2, and OpenAI says Calif found a critical flaw using GPT-5.6 Sol.

  • PHP released 8.5.12, 8.4.27, 8.3.36 and 8.2.35 on 21 October 2026 to fix CVE-2026-109618 and CVE-2026-109619, which can allow unauthenticated remote code execution.
  • PHP rates both flaws CVSS v4.0 9.2. Its advisory says updating is the only complete mitigation, and the mitigations it lists do not make an installation safe.
  • OpenAI says Calif identified a critical flaw using GPT-5.6 Sol, and that Wiz independently discovered and reported both vulnerabilities.
Checked 23 Oct, 09:20 · PHP project · 16 sources
Illustration: A patch panel with rows of ports; its green plug-in module, held just outside an open bay, stands for OpenAI's models that are coming to Google Cloud's Model Garden. Illustration made with AI for ai notis

Compute · 23 OCT

OpenAI will bring its models and Codex to Google Cloud

OpenAI announced the partnership with Google Cloud on 21 October, and says general availability is planned in the coming weeks.

  • OpenAI says it and Google Cloud are bringing OpenAI's frontier models and Codex agentic workflows into Google Cloud's Model Garden, in a post of 21 October 2026.
  • OpenAI says customers will buy the models through the Google Cloud Marketplace, and enterprises with Google Cloud spend commitments will be able to draw those down for the purchases.
  • OpenAI says general availability is planned in the coming weeks and its models will soon be in Google Antigravity and the Gemini Agent. Microsoft remains its primary cloud partner.
Checked 23 Oct, 04:29 · OpenAI · 1 source
Illustration: An archive cabinet with a drawer of identical folders; its green handover plate stands for ownership of the standard passing to a foundation. Illustration made with AI for ai notis

Tooling · 23 OCT

Anthropic donates Agent Skills to the Agentic AI Foundation

Anthropic's post of 22 October says the name, logo, GitHub organization and domain move to the foundation, which operates under the Linux Foundation.

  • Anthropic says it is donating the Agent Skills name and logo, the GitHub organization and the agentskills.io domain to the Agentic AI Foundation, which operates under the Linux Foundation.
  • Anthropic says more than 70 agent platforms, including ChatGPT, Cursor, GitHub Copilot and VS Code, have integrated Agent Skills, and nearly 280 million skills have been installed.
  • The proposed roadmap includes versioned releases, a formal proposal process and security guidance for skill authors and installers. Engineers from several companies will maintain the standard as individuals.
Checked 23 Oct, 04:29 · Anthropic · 1 source

You're all caught up

That's today's news: three stories.

ExplainedHow an agent loads a skill in three steps

You can join ai notis soon, to follow cases and keep stories on every device.

Follow us on Instagram and Threads

The morning email

On the mornings we publish: the three top stories and up to four short ones. Free.

We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep

No story today matches. Open the archive
Today's stories 3 STORIES

Today's stories

Fri 23 Oct · 3 STORIES
  1. Safety · 23 OCT · narrowed

    PHP patches two critical remote code execution flaws

    Illustration: A sealed junction box with a hinged service door; its green latch stands for the patch that closes both flaws.Illustration made with AI for ai notis

    The PHP project released updates on 21 October for two flaws rated CVSS 9.2, and OpenAI says Calif found a critical flaw using GPT-5.6 Sol.

    Checked 23 Oct, 09:20 · PHP project · 16 sourcesRead

  2. Compute · 23 OCT

    OpenAI will bring its models and Codex to Google Cloud

    Illustration: A patch panel with rows of ports; its green plug-in module, held just outside an open bay, stands for OpenAI's models that are coming to Google Cloud's Model Garden.Illustration made with AI for ai notis

    OpenAI announced the partnership with Google Cloud on 21 October, and says general availability is planned in the coming weeks.

    Checked 23 Oct, 04:29 · OpenAI · 1 sourceRead

  3. Tooling · 23 OCT

    Anthropic donates Agent Skills to the Agentic AI Foundation

    Illustration: An archive cabinet with a drawer of identical folders; its green handover plate stands for ownership of the standard passing to a foundation.Illustration made with AI for ai notis

    Anthropic's post of 22 October says the name, logo, GitHub organization and domain move to the foundation, which operates under the Linux Foundation.

    Checked 23 Oct, 04:29 · Anthropic · 1 sourceRead

toned down Safety 23 OCT

PHP patches two critical remote code execution flaws

The PHP project released updates on 21 October for two flaws rated CVSS 9.2, and OpenAI says Calif found a critical flaw using GPT-5.6 Sol.

PHP project · 16 sources·checked 23 Oct, 09:20
CalifGPTOpenAIPHPWiz
Illustration: A sealed junction box with a hinged service door; its green latch stands for the patch that closes both flaws.Illustration made with AI for ai notis
Compute

OpenAI will bring its models and Codex to Google Cloud

OpenAI announced the partnership with Google Cloud on 21 October, and says general availability is planned in the coming weeks.

23 OCT · OpenAI · 1 source
Tooling

Anthropic donates Agent Skills to the Agentic AI Foundation

Anthropic's post of 22 October says the name, logo, GitHub organization and domain move to the foundation, which operates under the Linux Foundation.

23 OCT · Anthropic · 1 source

Explained

One idea behind the news
Tooling How an agent loads a skill in three steps A skill is a folder with one required SKILL.md file; the agent reads a short description of each skill at start-up and opens the rest only when a task needs it.
In brief 4 more today
0 of 3 read
Drag a card to start Drag left or right · tap to read more Hover to lift a card · tap to bring it forward Tap again to put it back Tap a card to open it Tap again to close it Tap a headline to read it as a card