Episode Details

Please support this podcast by checking out our sponsors:
- Prezi: Create AI presentations fast - https://try.prezi.com/automated_daily
- KrispCall: Agentic Cloud Telephony - https://try.krispcall.com/tad
- SurveyMonkey, Using AI to surface insights faster and reduce manual analysis time - https://get.surveymonkey.com/tad


Support The Automated Daily directly:
Buy me a coffee: https://buymeacoffee.com/theautomateddaily

Today's topics:

Hidden AI reasoning traces leak - Security researchers claim hidden chain-of-thought can be recovered from encrypted API outputs from OpenAI, Anthropic, and Google models. The paper raises major concerns around AI privacy, reasoning trace leakage, distillation, and exposed secrets like API keys and internal URLs.

Compression and LLMs share math - A clear explainer connects data compression, entropy, arithmetic coding, and LLM prediction. It shows that compressors and language models both rely on next-token probability, making AI easier to understand through classic information theory.

Mojo 1.0 reaches stability - Mojo has reached version 1.0, its first stable release, with a stronger promise of compatibility for future development. That matters for developers watching high-performance languages for AI infrastructure, systems programming, and Python-adjacent workflows.

Text prompts become 3D worlds - Tencent's WorldClaw research generates large editable 3D environments from a single text prompt. The work is notable for combining open-world scale, local detail, and reusable assets, which could matter for games, simulation, and content creation.

Gorton, New York's working font - A detailed essay on the Gorton typeface traces how a plain industrial font spread across New York signs, keyboards, and equipment. It is a reminder that design history often hides inside everyday tools, manufacturing, and infrastructure.

-Researchers Find Hidden Reasoning Traces Can Be Recovered from LLM APIs
-Compression and LLMs Share the Same Prediction Problem
-Modular Announces Mojo 1.0 and MAX 26.5
-X.ai Launches Grok Bot as an AI Teammate
-Photosynthesis App Promises Two-Lens iPhone Photos With Real Optical Detail
-Satirical Tool Mocks LinkedIn Thought Leadership
-llama.cpp Promotes Private Local AI on Any Hardware
-aresluna.org
-Tencent Hunyuan3D Unveils WorldClaw for Editable 3D World Generation
-NVIDIA Launches Nemotron 3.5 Lightning and NeMo Switchyard


Episode Transcript

Hidden AI reasoning traces leak
We start with the most consequential story today: a security paper that challenges one of the core assumptions around modern reasoning models. The researchers say they were able to recover proprietary reasoning traces from