The model news is thin this Sunday. Four first-party posts set the day. They cover science access in Europe, election safeguards in the US, Nordic capacity planning, and agent quality at AWS. No new foundation model ships.
In the news
NVIDIA joins EC AI in Science Enablers and opens DLI courseware
NVIDIA posted on Oct 15 from the AI in Science Summit in Dublin. The European Commission launched the AI in Science Enablers Initiative tied to RAISE. NVIDIA pledged tools, compute, skills and open DLI courseware. The DLI hub is live and openly licensed. AI factory routes are named for France, Spain and Sweden. EuroHPC centers CINECA and JSC get GPU access for open research.1
Anthropic prepares for the 2026 US midterms
Anthropic published a policy post on Oct 14. It covers about 250,000 US Claude.ai conversations from Sep 21 to Sep 27. The TurboVote banner remains for registration and polling questions. An official voting information pull from Democracy Works is planned for later this month. The company reports 99 percent evenhandedness on Opus 5.5, Sonnet 5.5 and Haiku 5.5.2
NVIDIA Nordics 1 gigawatt plan and Sferical One
NVIDIA posted on Oct 14 about Nordic AI cloud growth. The company projects 1 gigawatts of capacity across the region by end 2027. Sferical One in Sweden is described with two DGX SuperPODs, 16 racks of GB300 and 1,152 GPUs. The consortium includes Ericsson, AstraZeneca, Saab, SEB, Wallenberg Investments and SKF. GTC Berlin keynote is Oct 21.3
AWS Bedrock AgentCore evaluation framework
AWS published an ML blog on Oct 15 about AgentCore. SMGS field sales built a multi-agent system on Bedrock AgentCore for worldwide sales. Pre-launch evaluation found a $410.08 revenue figure spoken as $410.08 Million. A coordinated release mismatch dropped pass rate from 78 percent to 38 percent. The post describes a nine-stage evaluation framework.4
