# News: October 15, 2026

> 4 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier.

- Author: [Médéric Hurier (Fmind)](https://www.fmind.dev/)
- Published: 2026-10-15
- URL: https://www.fmind.dev/news/2026-10-15/

- [Research@ 2026: Novel capabilities and intelligence](<https://blog.google/innovation-and-ai/technology/research/research-capabilities-intelligence-2026>): Google shares key breakthroughs from Research@ 2026, highlighting autonomous geospatial reasoning agents, Deep Review for technical document analysis, and planetary foundation models.
- [SemiAnalysis on Google TPUv7 Agentic Inference Benchmark](<https://x.com/SemiAnalysis_/status/2110560412907262343>): SemiAnalysis reveals Google's initial TPUv7 FP8 vLLM benchmarks deliver 1.67x better performance per dollar than Nvidia GB300 NVL72 FP4 on agentic inference workloads.
- [Anthropic flags revenue surge to investors ahead of stock market debut](<https://www.bbc.com/news/articles/cqzjx448vxkgo>): Anthropic reports a 45% quarterly revenue surge to $16.8 billion and its second consecutive profitable quarter as it prepares for an initial public offering.
- [AI agents aren't 'going rogue': Frontier labs are skipping security basics](<https://www.csoonline.com/article/4234854/ai-agents-arent-going-rogue-frontier-labs-are-skipping-security-basics.html>): Cybersecurity experts argue that recent AI agent security failures stem from inadequate vendor sandboxing and monitoring controls rather than rogue behavior, urging teams to tighten containment baselines.