# News: September 3, 2026

> 16 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier.

- Author: [Médéric Hurier (Fmind)](https://www.fmind.dev/)
- Published: 2026-09-03
- URL: https://www.fmind.dev/news/2026-09-03/

- [Introducing Gemini 3.8 Flash and 3.8 Flash Cyber](<https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber>): Google has released Gemini 3.8 Flash and 3.8 Flash Cyber, delivering next-generation intelligence optimized for agentic workflows and automated cybersecurity operations.
- [Introducing Muse Spark 1.3](<https://research.meta.ai/blog/introducing-muse-spark-1-3>): Meta has introduced Muse Spark 1.3, an upgraded foundation model engineered for complex reasoning and long-horizon agentic coding tasks across Muse Code and the Meta Model API.
- [Introducing Agentic Video Understanding with Gemini](<https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-agentic-video-in-gemini>): Google DeepMind has launched agentic video understanding across Gemini models, reducing token consumption by up to 88% and cutting inference costs by 66% while improving accuracy.
- [TabFM Adds Predictive ML to BigQuery](<https://cloud.google.com/blog/products/data-analytics/tabfm-adds-predictive-ml-to-bigquery>): Google Cloud has introduced TabFM in BigQuery preview, enabling zero-shot regression and classification on tabular datasets via SQL queries using in-context learning.
- [Databricks Adds Zhipu AI GLM 5.3 to Foundation Model APIs](<https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models#glm-5-3>): Databricks has expanded its Foundation Model APIs to support Zhipu AI's GLM 5.3 and GLM 5.3 Flash for pay-per-token and provisioned enterprise inference.
- [Getting Started with the Mantis Harness to Find and Fix Bugs](<https://cloud.google.com/blog/products/identity-security/getting-started-with-the-mantis-harness-to-find-and-fix-bugs>): Google Cloud has open-sourced the Mantis harness, an AI-driven framework to automate software vulnerability discovery, triage, reproduction, and remediation.
- [4 Engineering Patterns Behind the Strongest AI Agents Challenge Submissions](<https://developers.googleblog.com/4-engineering-patterns-behind-the-strongest-ai-agents-challenge-submissions>): Google Developers highlights four production architectural patterns from top AI Agents Challenge submissions, including bidirectional MCP communication and tiered intent routing.
- [Proactive Cyber Defense with the Fairwind Program](<https://blog.google/innovation-and-ai/technology/safety-security/fairwind-program>): Google has launched the Fairwind Program to provide governments and trusted enterprise partners with advanced AI cyber defense capabilities.
- [Claude Fable 5.1 Tops the Artificial Analysis Intelligence Index](<https://artificialanalysis.ai/articles/claude-fable-5-1>): Artificial Analysis reports that Anthropic's Claude Fable 5.1 scored an index-leading 66 at maximum reasoning effort, paired with a 75% price cut for cached input tokens.
- [Researchers Fear Safety Disaster Ahead of OpenAI’s Astra Release](<https://www.theverge.com/ai-artificial-intelligence/988334/openai-astra-ai-monitoring-safety>): The Verge reports on mounting concerns among AI safety researchers regarding governance and sandbox isolation protocols ahead of OpenAI's Astra release.
- [The Most Interesting Hack in History Just Got Weirder](<https://www.youtube.com/watch?v=0Rp9KJCEIvg>): Fireship analyzes OpenAI's official postmortem on the Hugging Face breach, exploring multi-agent exploit dynamics and container infrastructure vulnerabilities.
- [My 5 Predictions About the Future of Analytics](<https://www.youtube.com/watch?v=Inaon9qNAzE>): Colin Zima, co-founder and CEO of Omni, shares five industry predictions on the evolution of data teams, business intelligence dashboards, and natural language analytics.
- [Try Google Pics: Easy Image Creation and Editing in Google Workspace](<https://blog.google/products-and-platforms/products/workspace/google-pics>): Google has launched Google Pics, an AI image creation and editing tool powered by Nano Banana integrated into Google Docs, Slides, and Drive.
- [Automate Drive, Gmail, and Google Chat Actions with New Steps in Workspace Studio](<https://workspaceupdates.googleblog.com/2026/09/automate-drive-gmail-and-google-chat-actions-with-new-steps-in-Workspace-Studio.html>): Google Workspace Studio has introduced four new automation steps for file handling and messaging with configurable administrative approval controls.
- [Turn Google Docs, PDFs, and Word Files into Video Summaries in Google Vids](<https://workspaceupdates.googleblog.com/2026/09/turn-google-docs-pdfs-and-word-files-into-video-summaries-in-Google-Vids.html>): Google Vids now automatically converts text documents, PDFs, and Word files into concise video summaries with AI-generated scripts, narration, and visuals.
- [Custom Instructions for Gemini in Workspace Now Available in More Apps](<https://workspaceupdates.googleblog.com/2026/09/custom-instructions-for-gemini-in-Workspace-now-available-in-more-apps.html>): Google Workspace has expanded persistent custom instructions for Gemini across additional apps to provide personalized styling and consistent formatting across conversations.