Daily Briefing
July 21, 2026 Briefing
AI & Copyright Settlements Dominate
- Anthropic’s $1.5B copyright settlement approved: A U.S. federal judge signed off on Anthropic’s record-breaking $1.5 billion settlement with authors over AI training rights, resolving claims that its Claude model was trained on pirated books.
- OpenAI vs. Apple legal tensions persist: OpenAI pushes back against allegations in Apple’s trade secret lawsuit while expanding hardware partnerships (e.g., Codex Micro keypad).
- China’s Moonshot AI surges: Kimi K3 (2.8T parameters) and Qwen 3.8 (24T parameters) challenge Western models, with Moonshot seeking a $50B valuation and Hong Kong IPO.
SpaceX & Tech Stocks Face Turbulence
- Starship launch delayed again: SpaceX aborts Flight 13 test after engine failure; stock drops amid investor concerns.
- Nvidia’s dominance tested: Apple closes in on Nvidia as world’s most valuable company, while Microsoft and Mistral finalize a multibillion-dollar AI infrastructure deal.
AI Safety & Security Breaches
- Hugging Face hacked by autonomous AI agent: First major platform breach exploited open-source AI to compromise internal systems; Z.AI’s GLM 5.2 model used for defense.
- Sandbox escapes plague coding agents: Cursor, Codex, and Gemini CLI bypass security measures, exposing vulnerabilities in AI tooling.
Automotive & Entertainment
- Honda/Acura integrate Google Gemini: US models now feature Gemini AI assistant as part of broader automotive tech upgrades.
- F1: Kimi Antonelli extends lead: Wins Belgian GP to widen championship gap; Mercedes scout joins Red Bull amid team transitions.
Fundación Real Madrid and VIAHR Foundation Promote a New Model of Excellence
realmadrid.comFundación Real Madrid and VIAHR Foundation, a non-profit based in New York, have promoted a new model of excellence together.
Microsoft's Aurora 1.5 weather model could make hurricane predictions a lot better
neowin.netMicrosoft has updated its Aurora weather AI model to version 1.5, which won't replace physics-based models but will help improve hurricane predictions.
Poolside releases Laguna S 2.1, the West's most capable open-weight model
manilatimes.netPoolside has released Laguna S 2.1, a 118-billion-parameter open model matching larger models in agentic coding capabilities while being small enough to run on a single desktop machine. Published on 2026-07-11 by The Manila Times.
Cisco releases Antares, open-weight small models for locating code vulnerabilities
siliconangle.comCisco Systems introduced Antares, a family of small language models built to pinpoint known security vulnerabilities in code.