Anthropic announces Claude Fable 5.2 and Mythos 5.2
Anthropic says Fable 5.2 and Mythos 5.2 improve coding, reasoning, and visual skills and support software and hardware tasks. Fable, built on the same underlying model, is open to all; cybersecurity or biomedical research organizations may apply for Mythos. The company says safety evaluations cannot fully predict real-world behavior.
Anthropic announced two Claude 5.2 models, saying they improve coding, reasoning, and visual skills; organizations can apply for access to Mythos.
02MODELS
Muse Ember: a new model for repetitive coding and agent tasks
Muse Ember was introduced as a new model that handles text, images, audio, and video, with a 1-million-token context window. The company lists API pricing at $0.10 per million input tokens and $0.50 per million output tokens; claims of superior performance and efficiency are based on the company’s evaluations.
The company released its small Muse Ember model for coding subtasks and repetitive work by agents.
03SECURITY
Gemini may expose Drive and Gmail content to human review
The author says Google’s updated help text and blog edit disclose that Drive and Gmail content connected to Gemini may receive human review. In the author’s tests, ordinary chats sometimes triggered automatic, probabilistic data retrieval from files, ranging from summaries to full-text transfers. These are the author’s test claims.
Google warns that Drive and Gmail content connected to Gemini may be reviewed by humans; the author says tests show ordinary conversations can also retrieve file data.
04SECURITY
Claude Code settings files can leak secrets on GitHub
Apiiro researchers found more than 100,000 instances of Claude Code local settings files uploaded to public GitHub repositories. According to Apiiro, about 1,100 contained API tokens, database passwords, or private keys belonging to 63 organizations. Secrets may remain in forks even after files are deleted from a repository.
Apiiro researchers reported that uploading Claude Code’s local settings files to GitHub can expose thousands of sensitive items.
05INDUSTRY
Disney and Adobe to develop generative AI models for Disney
Disney and Adobe formed a multiyear partnership to develop artist-focused generative AI models for Disney’s production workflows. Adobe will provide early technology access through Firefly Foundry; the companies say the models will be customized for Disney characters, settings, and visual styles.
Under a multiyear partnership, Adobe will give Disney early technology access, and the companies will develop generative AI models for Disney’s production workflows.
06PRODUCTS
A $49.99 add-on for AI use in Logic Pro
Apple added a one-time $49.99 “Intelligence Usage Pack” for people who exceed the usage limit of Logic Pro’s AI assistant. The pack can be used for some features based on cloud-hosted OpenAI or Google Gemini models; unused extra usage expires after 12 months and is tied only to the purchasing Apple Account.
Apple offers a one-time $49.99 usage pack for people who want to keep using Logic Pro’s limited AI features.
07MODELS
NVIDIA introduces GR00T 2 robot model at CoRL
At CoRL, NVIDIA introduced its open-world action model GR00T 2 and simulation tools for robot training. The company reports that across 50 tasks, the model achieved an average success rate of 71% from a single demonstration and 95% from five demonstrations.
At CoRL, NVIDIA announced GR00T 2 and new robotics tools for robots to learn and test skills in simulation.
08MODELS
User alleges Opus 5.5 reasoning changed on November 3
Melaugh wrote that he had Gemini 4 Argon analyze 1,400 Claude Code sessions and saw a sharp change in Opus 5.5’s reasoning on November 3. In his own debugging test, he said the “high” version failed to find a solution while “xhigh” succeeded; he stressed this was not a scientifically robust benchmark.
Shane Melaugh says he had 1,400 Claude Code sessions analyzed and alleges that Opus 5.5’s reasoning changed on November 3.
✓
That’s it for today.The rest of your time is yours.
Your daily technology digest, every morning in English. Free. Unsubscribe anytime.
Sources, in the open.
Stories are selected from Hacker News; accessible original articles are summarized and translated with AI. Summaries may contain errors; check the source beneath each story for details. Preparation starts daily at 06:55 Türkiye time (UTC+3), targeting publication before 07:00.
Stories are read aloud by an AI voice.
This edition covers articles shared on Hacker News on the date shown above.
Now on the App Store for iPhone
Your daily digest. Now in your pocket.
Read or listen to a new edition every morning with Kısa for iPhone, save stories for later and keep your streak.
