Skip to content

Archive

Week 40 · 2026

September 28 – October 4

176 events

Importance: monitoring Security 1 source

AI Snake Oil commentary questions arguments about existential AI risk

The authors of a commentary on the AI Snake Oil website acknowledge that warnings about existential AI risk can be sincere, but consider many of the arguments flawed. They support evidence-based preparation for catastrophic risks and question the benefit of extreme warnings for safety.

Importance: monitoring Anthropic Tools and apps ✓ official

Delayed crediting on the Claude Platform is resolved

The Claude Status status page marked the incident of delayed crediting of purchased credits as resolved on 1 October at 22:36 UTC. Due to the unupdated balance, some requests failed for lack of credit.

Importance: monitoring GitHub Coding 1 source

GitHub's new dashboard is now the default for all users

GitHub's new dashboard, previously available as a preview, has become the default view for all users. The previous view can be restored at any time.

Importance: monitoring Google Tools and apps update ✓ official

The Guided Vision feature is also available in Google TalkBack

The Guided Vision feature is also available in Google TalkBack. On devices running Android 9 and later, an accessibility shortcut can be set up. The audio description of images from the camera helps with reading text and recognizing objects; according to Google, it is not intended for navigation.

New: Guided Vision is also available in Google TalkBack.; On devices running Android 9 and later, an accessibility shortcut can be set up for Guided Vision.; Google warns against using the feature for navigation and obstacle detection.

Importance: monitoring Open-source ✓ official

Olmo-core 3 introduces redesigned open infrastructure for training large MoE models

Olmo-core 3 has been released, providing infrastructure that keeps experts in GPU memory and routes data to them when training MoE models. According to the authors, it achieved approximately 2.7 times the throughput of the previous implementation in a preliminary test on eight NVIDIA B300 GPUs.

Importance: monitoring Anthropic Tools and apps 1 source

Anthropic offers the Claude for Government product to US agencies in open beta

The Claude for Government product has been in open beta for US federal and state agencies since July. It runs in a FedRAMP High environment, usage-based payments have fixed caps, and sensitive actions require two-person approval.

Importance: monitoring Business and investment 1 source

Local opposition and political conflict halt Stratos AI data center project in Utah

According to The Verge, resident opposition and political conflict halted the Stratos project, also known as Wonder Valley. The planned AI complex in Utah was to cover 40 000 acres and have a capacity of nine gigawatts. Residents only learned of the plans at a public announcement.

Importance: monitoring Business and investment 1 source

Analysis: cheaper GLM model could pressure margins of Western AI companies

According to the article, the GLM model from the company Zhipu competes with the products Claude and ChatGPT at a fraction of the price. The author anticipates pressure from cheaper open models on the margins of Western AI companies. It does not state specific prices or the model version.

Importance: monitoring Regulation and law 1 source

US Senate subcommittee discusses developer liability for AI agent attacks

A US Senate subcommittee addressed AI agent attacks and the liability of their developers. The central example was an incident at Hugging Face. Attorney Paul Ohm stated that some existing laws may be applicable, but recommended clearer legislation.