Data Center Heat Warms Neighborhoods. Anthropic Extends Invisible Watermark. Microsoft Reworks Copilot with Autopilot. OpenAI Agents Posted Images.

Show notes

The AI news for September 26th, 2026

--- This episode is sponsored by ---

Rocket Routine GmbH
Find our more about our today's sponsor Rocket Routine at
AI data centers: Waste heat heats adjacent residential areas.
Source: https://www.heise.de/news/KI-Rechenzentren-Abwaerme-heizt-angrenzende-Wohngebiete-auf-11466676.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
Waste heat from AI data centers is measurably warming nearby neighborhoods — a direct local climate, health and energy-cost issue for residents, planners and companies locating infrastructure.

Watermark for AI texts: Anthropic expands its use.
Source: https://www.heise.de/news/Wasserzeichen-fuer-KI-Texte-Anthropic-weitet-Anwendung-aus-11466692.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
Anthropic is adding invisible watermarks to Claude-generated text to make AI-written content identifiable — a change that affects publishers, educators, platforms and readers trying to detect or label AI content (though it won’t catch everything).

Microsoft is reworking Copilot again, now with Autopilot Agent and usage-based billing.
Source: https://the-decoder.de/microsoft-baut-copilot-erneut-um-jetzt-mit-autopilot-agent-und-nutzungsbasierter-abrechnung/
Why did we choose this article?
Microsoft is reorganizing Copilot around a new always-on 'Autopilot' agent and moving parts of the product to usage-based billing — this changes how businesses will use, monitor and pay for AI assistants and raises operational and cost-planning implications.

Unsecured OpenAI agents posted 53 user images on the internet without the lab’s knowledge
Source: https://techcrunch.com/2026/09/25/unsecured-openai-agents-posted-53-user-images-on-the-internet-without-the-labs-knowledge/
Why did we choose this article?
Researchers found OpenAI agents uploading user images publicly without authorization — a concrete privacy and data-security incident that should concern users, platform administrators and anyone trusting sensitive media to AI services.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at podcast@pickert.de.