# News: October 9, 2026

> 12 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier.

- Author: [Médéric Hurier (Fmind)](https://www.fmind.dev/)
- Published: 2026-10-09
- URL: https://www.fmind.dev/news/2026-10-09/

- [Introducing the Gemini agent](<https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026>): Google Cloud announces the Gemini agent, a universal workplace agent featuring persistent cloud execution, sub-agent orchestration, and deep Workspace integration.
- [Introducing Personal Agent Protocol](<https://sierra.ai/blog/introducing-personal-agent-protocol>): Sierra and Meta introduce the Personal Agent Protocol with Shopify and Stripe, establishing an open OAuth-based standard for personal AI agents to securely interact with enterprise services.
- [Ambient Quality Agent (AQuA)](<https://developers.googleblog.com/the-outer-loop-insights-first-an-ambient-quality-agent-that-diagnoses-your-production-agent>): Google open-sources AQuA, an ambient diagnostic agent that continuously analyzes production agent traces, logs, and code snapshots outside the request path to cluster failure patterns.
- [Anthropic OSS Scanner](<https://www.anthropic.com/research/launching-opt-in-vuln-finding-service-for-open-source>): Anthropic launches OSS Scanner, an opt-in service using Claude frontier models to provide free vulnerability scans and coordinated disclosure for open-source projects.
- [GitHub Copilot CLI Downloads and Executes Malware](<https://www.promptarmor.com/resources/github-copilot-cli-downloads-and-executes-malware>): Prompt Armor reveals an indirect prompt injection vulnerability in GitHub Copilot CLI that bypasses command validation to download and execute arbitrary malware without user confirmation.
- [Scaling agentic coding practices across your team](<https://www.youtube.com/watch?v=TiH2E2mPZIk>): Google Cloud Tech demonstrates how engineering teams can scale agentic coding by defining structured markdown specification files to guide autonomous software agents across complex codebases.
- [Navigating AI automation in Looker: Essential developer tools for agents](<https://www.youtube.com/watch?v=JvASK8qX4hg>): Google Cloud Tech highlights five essential developer tools in the Agentic Looker Developer Stack, including Looker Skills, CLI, and managed MCP servers to automate data modeling and LookML workflows.
- [Anthropic Coordinated Vulnerability Disclosure Dashboard](<https://red.anthropic.com/2026/cvd>): Anthropic unveils its public CVD dashboard, documenting over 6,157 open-source vulnerabilities uncovered by Claude models and validated by independent security researchers.
- [2026 Usage Policy update](<https://www.anthropic.com/news/2026-usage-policy-update>): Anthropic updates its Usage Policy to clarify compliance requirements for autonomous agents, physical hardware controls, and elections while consolidating rules against deceptive influence operations.
- [OpenAI annualized revenue falls $20B below prior estimates](<https://finance.yahoo.com/technology/article/openais-annualized-revenue-20-billion-lower-than-prior-investor-estimates-174058048.html>): The Financial Times reports OpenAI's annualized revenue reached ~$50 billion in September, falling $20 billion below prior investor estimates and trailing Anthropic's $65 billion run rate.
- [Gemini at Work 2026](<https://www.youtube.com/watch?v=1ASzWklab2U>): Google Cloud CEO Thomas Kurian delivers the keynote at Gemini at Work 2026, highlighting global customer transformations and the enterprise rollout of the universal Gemini agent.
- [Mathieu on X](<https://x.com/miniapeur/status/2108124820357460200>): Mathieu highlights research showing how AI autoformalization can mistranslate natural language proofs into Lean, producing compiling verifications like OpenAI's Navier-Stokes proof that depart from the original claims.