Disney Sues Midjourney for Copyright Infringement

Also, Mistral eyeing major growth as Europe appears fearful of US AI dominance

Author

Jack Lajoie
June 13, 2025 • read time ~ 9 minutes

⚡️ Headlines

🤖 AI

Microsoft Copilot zero‑click vulnerability “EchoLeak” patched – A critical flaw (CVE‑2025‑32711) in Microsoft 365 Copilot allowed data exfiltration without user interaction, prompting an urgent patch and raising concerns about AI agent security. [Fortune]

Meet Green Dot Assist: Starbucks generative‑AI‑powered coffeehouse companion – Starbucks launched an AI assistant for baristas, hosted on iPads, offering drink recipes, equipment troubleshooting, staffing support, and real-time guidance; currently piloting at ~35 stores. [Starbucks]

Nvidia teams up with Mistral to build European AI cloud – Nvidia partners with French startup Mistral to create a local AI cloud infrastructure across Europe, reinforcing sovereign AI efforts. [Bloomberg]

Meta launches AI ‘world model’ to advance robotics and autonomy – Meta released V‑JEPA 2, a “world model” AI that enables robots and self‑driving systems to simulate and reason about 3D environments via unlabelled video. [CNBC]

CoreWeave to offer compute capacity in Google‑OpenAI cloud partnership – CoreWeave secured a role in Google’s upcoming OpenAI cloud deal, supplying essential GPU-based compute power. [Reuters]

Trump’s “Made in the USA” moment won’t be the iPhone—it’s Nvidia H100 chips – Reports indicate that US-made AI server components like Nvidia’s H100 chips, supported by firms like Foxconn and Wistron, may mark Trump’s first effective onshoring of high-tech manufacturing. [Nikkei Asia]

Meta hires top researchers from Google and Sesame for new AI lab – Meta lured prominent AI researchers such as Jack Rae (DeepMind) and Johan Schalkwyk (Sesame) to lead a high-budget “superintelligence” initiative. [Bloomberg]

Amazon debuts AI‑video‑generator for U.S. ad sellers – Amazon’s AI Video Generator tool is now available to U.S. advertisers, producing dynamic 21‑second ads from static images or existing footage. [The Verge]

Mattel partners with OpenAI to create its first AI‑powered toy – Mattel is collaborating with OpenAI to launch its first AI-enabled toy by late 2025, while adopting ChatGPT Enterprise internally. [Reuters]

🦾 Emerging Tech

Tesla robotaxi rides in Austin to begin June 22, per Elon Musk – Elon Musk confirmed robotaxi service will begin in Austin, Texas, around June 22, 2025. [CNBC]

Shopify to support USDC payments on Coinbase’s Base for global merchants – Shopify will integrate USDC stablecoin payments via Coinbase’s Base network, expanding crypto options for merchants worldwide. [CoinDesk]

🤳 Social Media

Meta launches AI-powered video editing with preset themes – New AI-powered presets (e.g., desert, anime) in Meta’s video editing tools allow automatic styling of the first 10 seconds of uploads in select countries. [The Verge]

Brazil’s top court moves to hold platforms liable for user content – Brazil’s Supreme Court approved holding platforms accountable with fines for failing to proactively remove illegal content. [Reuters]

Instagram’s new ‘creator inspiration’ campaign includes grid rearranging – Instagram launched features allowing users to rearrange their profile grid and explore creative prompts under its new Creator Inspiration campaign. [Social Media Today]

🔬 Research

Brazil uses AI for Amazon-region healthcare prescriptions – Brazilian healthcare providers are piloting AI tools to manage and personalize prescriptions in the Amazon region. [Rest of World]

Unprecedented molecular simulation dataset opens AI research – Researchers released a massive molecular simulation dataset to accelerate AI-driven drug discovery efforts. [Phys.org]

⚖ Legal

EU court rules on Meta and TikTok platform fees – The EU’s second‑highest court weighed in on tech-platform fee structures involving Meta and TikTok, detailing implications for fee compliance. [Reuters]

UK police criticized over facial recognition data-sharing – UK police forces were reprimanded for sharing facial recognition data with commercial firms, sparking a debate over privacy and oversight. [BBC]

🎱 Random

Adobe posts strong outlook touting traction for AI tools – Adobe reported a bullish financial forecast, crediting steady growth in adoption of its AI-powered Creative Cloud features. [Bloomberg]

🔌 Plug-Into-This

Disney and NBCUniversal have filed a landmark copyright infringement lawsuit against Midjourney, accusing the AI platform of unauthorized use of their intellectual property in both training data and image outputs. The lawsuit, filed in California federal court, includes direct visual comparisons of AI-generated content and iconic studio characters, arguing Midjourney has ignored prior warnings.

  • Targets both direct and secondary infringement by highlighting how Midjourney trains on and reproduces proprietary IP.

  • Lawsuit includes side-by-side examples of AI outputs mimicking characters like Spider-Man and Shrek.

  • Studios allege Midjourney refused to adjust its practices, unlike peers who responded to takedown demands.

  • Legal strategy emphasizes platform responsibility over individual user behavior.

  • Could set precedent requiring AI companies to source licensed training data or limit model capabilities.

👨‍⚖️ This legal move signals an industry pivot—from tolerating downstream misuse to targeting upstream AI tools—underscoring how foundational training data choices may now carry real legal and financial risk.

French startup Mistral is rapidly gaining international traction as geopolitical concerns about U.S. tech dominance push European and global clients toward regionally controlled AI providers. The company is building a massive Nvidia-powered data center outside Paris and is reportedly considering raising another $1 billion to scale further.

  • Plans a 40 MW data center—expandable to 100 MW—powered by around 18,000 Nvidia GPUs.

  • More than half of its projected $100 million revenue is expected to come from non-European clients.

  • Mistral’s open-weight models are seen as competitive with U.S. leaders in performance benchmarks.

  • National governments are directly supporting its growth under “sovereign AI” initiatives.

  • The firm is preparing for further capital raises amid rising demand from non-U.S. entities.

🌍 Mistral's momentum reveals a broader trend: AI infrastructure is becoming a proxy for national sovereignty, and the rise of regionally aligned players may reshape where innovation—and influence—happens.

The Browser Company has launched Dia, a browser that embeds AI deeply into everyday web use. Unlike conventional assistants or plugins, Dia integrates AI into its core UI and memory—leveraging user sessions, cookies, and task-specific tools to perform personalized actions directly within the browser.

  • Introduces an omnibox that can handle search, navigation, and AI prompts interchangeably.

  • Uses modular “skills” to route tasks—like summarizing, shopping, or writing—to specialized AI models.

  • Prioritizes model-routing intelligence over building its own foundational LLM.

  • Accesses session data and cookies to offer context-aware support across tabs and sites.

  • Implements privacy-conscious architecture with local encryption and minimal cloud storage.

🔎 Dia exemplifies the new direction for browsers: less about static windows, more about AI-enhanced interaction layers—turning the browser itself into a task-aware, memory-equipped digital aide.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse