AI News · October 23, 2026

This week in AI — ChatGPT ads go to 26 more countries, Agent Skills gets a neutral home

Official announcements only, Oct 18–23, 2026. ChatGPT ads expand, Anthropic hands the Agent Skills standard to a Linux Foundation project, OpenAI launches a program to patch the bugs AI agents are finding, and OpenAI models land on Google Cloud.

Key takeaways

From official announcements, Oct 18–23, 2026: OpenAI expanded ChatGPT ads to 26 more countries. Anthropic donated the Agent Skills standard to the Agentic AI Foundation, hosted by the Linux Foundation. OpenAI launched Patch the Planet to fix vulnerabilities found by AI agents, and said its models are coming to Google Cloud's Model Garden.

We only cover what companies published on their own blogs and newsrooms, and every item links to the original. Performance claims below are the companies' own, not independent tests.

One note this week: OpenAI's pages blocked our reader, so for OpenAI items we stick to the summaries in OpenAI's official RSS feed and go no further.

1. ChatGPT ads expand to 26 more countries (Oct 21)

OpenAI is bringing ChatGPT ads to 26 new countries. Per the headline, over 100,000 advertisers have used the platform so far.

The pitch to businesses: reach people while they're using ChatGPT for shopping research, whatever the size of the business.
How it works: a companion engineering post describes an LLM-based ad system that reads the intent of a conversation while keeping ads separate from answers.
Which countries: we couldn't open the full post, so we can't list them.

What it means for you: If you ask ChatGPT for product advice, expect to see more ads next to the answer. Keep the two apart in your head.

Source: ChatGPT Ads expands to 26 new countries — OpenAI · How we built ChatGPT's ad stack — OpenAI

2. Anthropic donates Agent Skills to the Agentic AI Foundation (Oct 22)

Agent Skills is the open standard for teaching AI agents a specific task or domain — a skill is just a folder with a SKILL.md file in it. Anthropic is handing it to the Agentic AI Foundation (AAIF), hosted by the Linux Foundation, along with the name, logo, GitHub organization and the agentskills.io domain.

Adoption: more than 70 agent platforms have picked it up since December 2025, including ChatGPT, Cursor, GitHub Copilot and VS Code. Anthropic cites 1 million published skills and nearly 280 million installs.
Who runs it: engineers from Anthropic, Google, OpenAI and others, maintaining it as individuals.
What's next: versioned releases, a formal proposal process, and security guidance for writing and installing skills.
You can keep using or contributing to the standard without joining the AAIF.

What it means for you: A skill you write or install should work in more tools than just Claude. Until the security guidance lands, be picky about installing skills from strangers.

Source: Donating Agent Skills to the Agentic AI Foundation — Anthropic

3. "Patch the Planet": fixing the bugs AI agents are finding (Oct 21)

As AI agents uncover critical vulnerabilities in open-source software, OpenAI launched a program that connects security researchers, project maintainers and incident responders to get those holes fixed.

Same day, OpenAI also published a policy blueprint for putting frontier AI in the hands of cyber defenders, and a case study on Cloudflare using OpenAI's tools to investigate vulnerabilities and propose fixes for review.

What it means for you: Faster discovery means more security updates. Turning on automatic updates for your phone, computer and apps is the easiest way to benefit.

Source: Patch the Planet — OpenAI

4. OpenAI models are coming to Google Cloud (Oct 21)

OpenAI's frontier models and Codex workflows will be available in Model Garden on Google Cloud, through Google Cloud Marketplace.

What it means for you: Nothing changes in your ChatGPT app. If your company runs on Google Cloud, OpenAI models become another option at work.

Source: OpenAI models are coming to Google Cloud — OpenAI

5. ChatGPT gets better at less common programming languages (Oct 19)

OpenAI says it's improving ChatGPT's coding in low-resource programming languages, with new evaluations covering software engineering, SQLite and COBOL migration.

What it means for you: If you maintain legacy code, it's worth another try. The progress is measured by OpenAI's own evaluations.

Source: Advancing coding in low-resource programming languages — OpenAI

In case you missed it

Anthropic Public Benefit Fund (Oct 18): Subject to stockholder approval, 5% of Anthropic's shares go into a new fund for the public good, with more added as the company's value grows. With earlier commitments, Anthropic says roughly 20% of its shares will be dedicated to the public good. The fund's equity won't be used for lobbying or political giving. Source — Anthropic
Tougher distillation defenses (Oct 12): Anthropic stepped up defenses against covert efforts to copy Claude's capabilities into other models. It estimates distillers now harvest 90–99% less data, with fewer than 0.1% of benign requests blocked by mistake. Some API developers had to update their integrations. Source — Anthropic

The takeaway

ChatGPT is becoming an ad platform, agent tooling is moving toward shared standards, and the security world is organizing around the bugs AI agents are finding. More tools only help if you use them well — find out your AI type in 3 minutes.

Sources

#AI news#ChatGPT#Claude#agents#security#ads

Written with AI (Claude) and checked against the linked sources.