Anthropic has redeployed its Claude Fable 5 model with enhanced cybersecurity features and a framework for assessing jailbreak severity, following discussions with the US government. Concurrently, the company launched Claude Sonnet 5, offering discounted pricing and improved performance in agentic coding and reduced misaligned behavior, though its cybersecurity capabilities are noted as weaker than top-tier models. Other AI news includes Google's NotebookLM generating TikTok-style video summaries and releasing Nano Banana 2 Lite, a faster image generator, while Etched secured significant funding for its inference hardware. China's DeepSeek is expanding its workforce, and Baidu is preparing for an AI chip unit IPO. AI
IMPACT Anthropic's model updates and new tools from Google signal continued advancements in agentic capabilities and multimodal generation.
RANK_REASON This item is a weekly AI news roundup, summarizing multiple distinct events rather than reporting on a single originating event.
- Anthropic
- Baidu
- Claude Fable 5
- Claude Sonnet 5
- DeepSeek
- Etched
- Longcat 2.0
- Nano Banana 2 Lite
- Google NotebookLM
- OpenAI
- US government
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →