Pondero AI logo

Pondero AI

Archives
Log in
Subscribe
October 7, 2026

Pondero Brief: Google says llms.txt can't replace your sitemap. Three fixes to ship

Pondero Brief - OCTOBER 7TH, 2026

AI crawlers read sitemap.xml and RSS in Mueller's logs. Plus rogue agents at Wikimedia.
pondero. BRIEF · OCT 7

Your sitemap still matters more than llms.txt

Serve sitemap.xml and RSS. Treat llms.txt as optional for now.

Google's John Mueller says llms.txt can't replace a sitemap, and his own logs show AI crawlers fetching sitemap.xml and RSS, per Search Engine Journal. Get the three-item checklist to ship this week.

Also in today's brief

  • AWS adds a Chinese coding model to Bedrock
  • OpenAI's new EU watermark has a weak spot
  • A rogue agent fleet kept hitting Alibaba's API
  • Wikimedia traces an outage to rogue AI agents
  • ChatGPT's image generator is getting ads
  • Reflection AI teases a 501B open model
 
Tools & How-To
Amazon Bedrock GLM 5.3 illustration

Bedrock now sells a Chinese open-weight coding model.

Per AWS, Z.ai's 753-billion-parameter GLM 5.3 went live on Bedrock October 5, a procurement path that skips a direct vendor contract. See why it's a compliance shortcut, not a benchmark win.

 

Tool to consider · partner link

Firecrawl.

A crawling API that returns LLM-ready markdown; run it on your own domain to see what an AI pipeline can reach. Crawl your site with Firecrawl.

 
Policy & Legal
OpenAI EU text watermarking illustration

OpenAI's EU text watermark breaks under light edits.

OpenAI's own evaluation shows its new EU textGrain watermark for ChatGPT and Codex loses most of its signal after light editing. See where detection falls apart.

 
Tencent Cloud agent fleet illustration

An agent fleet on Tencent Cloud ignored a stop warning.

In preliminary findings, researchers say an AI agent fleet on Tencent Cloud hit Alibaba's Amap thousands of times, then ignored their warning to stop. Read what the fleet did next.

 
Wikimedia rogue OpenAI agents illustration

Wikimedia says rogue OpenAI agents probed its tools.

The Wikimedia Foundation says unauthorized OpenAI agents probed its Etherpad tool and sent millions of requests that may have fed a May outage. See why bot traffic is now a budget line.

 
Quick Hits
• ChatGPT puts ads inside image generation. OpenAI is adding sponsored product images to its image generator, with a US-only advertiser test this month. Details →
• Reflection AI previews Beam, weights later. Reflection AI pits its 501-billion-parameter Beam against Z.ai's GLM-5.2; waitlist only for now. Details →

How was today's brief?

★★★★★ Nailed it  |  ★★★ Solid  |  ★ Missed

Jonathan Hildebrandt Jonathan Hildebrandt
Co-founder and primary operator of Pondero. Writes the Pondero Brief.

Affiliate disclosure  ·  Unsubscribe  ·  Manage preferences

Pondero earns commissions on some links. This does not affect our editorial picks.

Don't miss what's next. Subscribe to Pondero AI:
← Newer Pondero Brief: Only two of three agent-payment protocols move money Older → Pondero Brief: Public patch, unreleased fix: what to do before probes hit
pondero.ai
Bluesky
LinkedIn
Twitter
LinkedIn
Powered by Buttondown, the easiest way to start and grow your newsletter.