Section 1 : Hors podcast
- Google limits Meta’s use of its Gemini AI models, FT reports
- Robots humanoïdes - On attend vraiment qu’il y ait un mort ?
- cancel_claude
- Everything Is Logarithms
- APERTVS.ai
- Fine Tuning a Local LLM to Categorize Questions
- You should use AI for reviewing code especially when the diff is huge – Simian Words
- The text in Claude Code’s “Extended Thinking” output is not authentic. – blog
- Whose Agent Are You? Multi-Layer Fingerprinting and Attribution of Autonomous Web Agents
- Will It Mythos? - I’ve done some things
- How to burst the AI bubble: Strike at its roots
- AI Hiring Tools Can Yield Racial Bias and Systemic Rejection
- Open-source LLMs administer maximum electric shocks in a Milgram-like obedience experiment
- Legal Reasoning Is Not Lawyering: Rethinking Legal Benchmarks for Pro Se Access to Justice
- A global log for medical AI
- Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluations
- Detecting AI Coding Agents in Open Source: A Validated Multi-Method Census of 180 Million Repositories
- Governed Shared Memory for Multi-Agent LLM Systems
- Scaling Laws for Task-Specific LLM Distillation
- LLMs are Bayesian in Expectation, Not Realization
- Everyone is Wrong About AI Except Me
- Hotter Than a Hot Tub: The 45°C Breakthrough to Cool AI’s Biggest Machines
- Why Frontier AI makes prioritization the most important part of your CTEM program
- For Most of the World, Open-Source AI Is the Only Way Forward - Techstrong.ai
- A quote from Tom MacWright
- Walmart’s First Nuclear Deal Shows Demand Beyond AI Data Centers - Slashdot
- Entropy-Based Observability for AI Agent Behavior
- The Hitchhiker’s Guide to Agentic AI: From Foundations to Systems
- Agentic System as Compressor: Quantifying System Intelligence in Bits
- Political bias in AI · Where the AI models stand
- Anthropic says Alibaba must be punished for largest Claude cloning attack
- The Doorman Fallacy In Action
- Pannes d’IA - Les coupures ont été multipliées par 8 en un an
- AI and Liability
- Anthropic is testing desktop-like Claude Cowork for mobile
- Dream Server - Un serveur IA complet chez vous en une commande
- Why software engineers are grieving – Jim Grey on software management
- A quote from OpenAI
- DeepSpec/DSpark_paper.pdf at main · deepseek-ai/DeepSpec
- Forget Prompt Engineering: ‘Loop Engineering’ Is All the Rage Now - Slashdot
- The best response to AI slop, infinite advice, and online noise is from Robin Williams
- Ford hired AI and sacked humans. It backfired badly
- When Compression Becomes an Attack Surface: Black-Box Attacks on Prompt-Compressed LLM Agents
- Affective AI Safety: The Missing Piece in LLM Safety
- OpenAI leans toward waiting until next year for IPO, NYT reports
- Exploiting Neural Audio Codec Latents for Adversarial Audio Attacks
Section 2 : Podcast 100% humain
- Mythos…
- Anthropic’s Mythos AI Model Reportedly Breached NSA Classified Systems in Hours
- How Anthropic may have talked itself into an AI export ban
- Anthropic Confirms Claude Mythos 5 Redeployment for US Critical Infrastructure Organizations
- Mythos discovers ‘Squidbleed,’ a memory leak that’s gone undetected since Clinton era
- Chinese cybersecurity company claims it’s built a better-than-Mythos bug finder
- Asian AI startups launch Mythos-like models as Anthropic’s export ban drags on
- Surviving the Mythos Era: Richard Bejtlich on the Case for NDR
- Cybersecurity in the post-mythos era: Keep calm and carry on!
- Plus ça change
- Intel agencies: Frontier AI models will reshape cybersecurity faster than expected
- OpenAI Releases GPT‑5.5‑Cyber With Full Automation for Vulnerability Detection and Patching
- It’s looking like a hot, messy summer for security teams as AI finds countless previously hidden vulns
- Vulnerability Reports Are Not Special Anymore
- How Your Credentials Are Leaked by LLM Agent Skills: An Empirical Study
- Defense Effectiveness Across Architectural Layers: A Mechanistic Evaluation of Persistent Memory Attacks on Stateful LLM Agents
- Amplify, Don’t Create: Temporal Accumulation for Slow-Burn Prompt Injection
- Think Twice Before You Act: Protecting LLM Agents Against Tool Description Poisoning via Isolated Planning
- Honeyquest for LLMs: Rethinking Cyber Deception for AI Attackers
- Microsoft uses AI to link two malware operations in racketeering suit
- More Malicious OpenClaw Skills Threaten AI Supply Chain
- Fake AI Agent Skill Passed Security Scans and Reportedly Reached 26,000 Agents
- FALCON: Transforming Cyber Threat Intelligence into Deployable IDS Rules with Self-Reflection
- Open-source security is posing challenges governments can’t easily solve
- What happened after 2,000 people tried to hack my AI assistant