DeepSeek's AI faces censorship in China; OpenAI aims for ChatGPT as a 'super assistant'
Key Takeaways
- DeepSeek's R1 0528 model faces censorship concerns, particularly in China.
- OpenAI plans for ChatGPT to become a 'super assistant' for users.
- Google's Gemini will automatically summarize emails for users.
- ChatGPT reached 800M users in 17 months; Nvidia's Blackwell GPU uses less energy.
- TechCrunch Sessions: AI will be held on June 5, 2025, at UC Berkeley.
Top Stories
DeepSeek R1 model faces censorship concerns in China.
On May 30, 2025, Ryan Daws reported that DeepSeek’s latest AI model, R1 0528, is facing criticism for apparent censorship, particularly in China, which is considered a setback for free speech.
OpenAI aims for ChatGPT to be a 'super assistant'.
According to an internal strategy document published on May 30, 2025, OpenAI plans for ChatGPT to evolve into a "super assistant" for users, acting as their primary interface to the internet.
Google's Gemini to summarize emails automatically.
On May 30, 2025, Google announced that its AI assistant, Gemini, will now feature email summary cards, automatically condensing email content for users without requiring any interaction.
Mary Meeker's report: AI development, adoption, and usage.
On May 30, 2025, Mary Meeker released a 340-page report titled “Trends — Artificial Intelligence,” detailing the rapid development, adoption, spending, and usage of AI. The report notes that ChatGPT reached 800 million users in 17 months, Nvidia's 2024 Blackwell GPU uses significantly less energy, and companies like Google and Amazon are developing chips at scale for their clouds.
TechCrunch Sessions: AI on June 5, 2025.
TechCrunch Sessions: AI will be held on June 5, 2025, at UC Berkeley’s Zellerbach Hall, featuring speakers from Anthropic, Google Cloud, and OpenAI. Attendees can save on passes, and TechCrunch is hosting a trivia contest for a chance to win tickets.
Business AI Impact
TechCrunch Disrupt 2025 breakout session finalists announced.
On May 30, 2025, TechCrunch announced the finalists for breakout sessions at TechCrunch Disrupt 2025, including Anmol Rastogi from Amazon, Dan Hendrycks from CAIS, Jack Hidary and Jim Breyer, Matt Madrigal from Pinterest, Michael Stewart from M12, and Sridhar Ramaswamy from Snowflake.
Trending AI Tech
Advancing AI inference capabilities: hardware, ML, integration.
On May 30, 2025, MIT Technology Review Insights, in partnership with Arm, published an article discussing how advancements in inference capabilities can deliver accessible AI by focusing on hardware and silicon design, machine learning, and system integration. The article also notes Moore’s Law is plateauing, while Lightmatter is improving speed and efficiency with photonic computing, DeepSeek's R1 achieves reasoning with fewer resources, and Anthropic and Arm are setting industry standards for AI development.