Saturday, 4 July 2026
Memora cuts agent memory token use by 98%
Archive item — written before sources were shown.
Microsoft's open-source Memora compresses agent memory into brief abstractions, cutting context token use by 98% and setting new highs on two benchmarks.
Nvidia trades GPU financing for revenue cut
Archive item — written before sources were shown.
Nvidia is financing GPU buildouts for smaller cloud operators in exchange for a share of their cloud revenue; first partners are in Australia and Indonesia.
Pentagon demanded Anthropic enable weapons
Archive item — written before sources were shown.
Unsealed court emails show the Pentagon demanded Anthropic remove guardrails on autonomous weapons and domestic surveillance; Amodei refused.
Anthropic tracking code triggers Alibaba ban
Archive item — written before sources were shown.
Anthropic's Claude Code quietly flagged Chinese-timezone users via altered date and apostrophe characters; Alibaba bans the tool from July 10 as a result.
White House AI security EO hits first deadline
Archive item — written before sources were shown.
The June 2 AI security EO's frontier-model vetting framework was finished by its August 1 deadline, briefed to major labs — but its contents stay confidential.
Talp: AI customer personas, $20M pre-seed
Archive item — written before sources were shown.
Talp simulates customer reactions to products and pricing using AI personas; it closed at a $20M pre-seed valuation backed by Formus Capital and a16z Scout.
UN and ITU launch AI governance commission
Archive item — written before sources were shown.
The UN and ITU launched the AI for Good Global Commission on July 1, co-chaired by Marc Benioff and Paul Kagame, with Jensen Huang, Andy Jassy, and Brad Smith.
Global VC hit $510B in H1 2026, driven by AI
Archive item — written before sources were shown.
Crunchbase reports global venture funding reached a record $510B in H1 2026, more than all of 2025; OpenAI and Anthropic alone accounted for $217B of the total.
Stathera raises $55M for AI cluster timing
Archive item — written before sources were shown.
Stathera closed a $55M Series B led by Maverick Silicon to scale MEMS-based oscillator chips that improve GPU cluster synchronization in AI data centers.
Meta Watermelon matches GPT-5.5, leader says
Archive item — written before sources were shown.
Meta Chief AI Officer Alexandr Wang told employees Watermelon has caught up to GPT-5.5 on internal benchmarks; no release date or external results disclosed.
Cline 3.0.37 handles malformed tool calls
Archive item — written before sources were shown.
Cline v3.0.37 (July 4) gracefully handles malformed tool calls from weaker models like DeepSeek and ensures mode-switching notices reach the model.