Jun 13, 2026

• US Government directive to suspend access to Fable 5 and Mythos 5

  • The US government has mandated the suspension of access to Fable 5 and Mythos 5 for all foreign nationals, including Anthropic employees, due to concerns over a potential jailbreak method that could exploit minor vulnerabilities, despite no evidence of a universal jailbreak being found.

• US bans differential privacy in Census data

  • The U.S. Department of Commerce has banned noise infusion in statistical products, which undermines the effectiveness of privacy-preserving techniques like differential privacy that balance data utility and confidentiality.

• There is a shadow hanging over this Fable thing

  • The US government has mandated Anthropic to disable access to Fable and Mythos for all foreign nationals, citing national security concerns over potential vulnerabilities, which Anthropic argues are not unique to their models.

• AI OSS tool repo goes archived over night after raising $7.3M Seed

  • TensorZero is an open-source LLMOps platform that integrates multiple functionalities, including a unified API for LLM access, observability for inferences, and automated optimization through its Autopilot feature, enhancing LLM performance across diverse tasks.

• Maxproof

  • MaxProof is a population-level test-time scaling framework that enhances mathematical proof capabilities, achieving scores of 35/42 on IMO 2025 and 36/42 on USAMO 2026, surpassing human gold-medal standards.

• /architect: Reduce Fable tokens by 80%, Fable orchestrates/reviews, Codex builds

  • architect-loop integrates Claude Fable and GPT-5.5 Codex to automate software design and research, enabling efficient project management without the need for API keys or additional costs.

• Shepherd's Dog: A Game by the Most Dangerous AI Model

  • Anthropic's latest AI model has successfully created a complete game titled "Shepherd's Dog" in one attempt, showcasing its advanced capabilities with a single 2,319 line index.html file and no dependencies.

• Adaptive PDFs

  • Adaptive PDFs allow humans to view formatted documents while enabling machines to extract clean markdown structure from the same file, enhancing accessibility for LLMs and text extractors.

• Our response to the US ban on Fable 5 and Mythos 5

  • The US export control directive banning access to Anthropic’s LLMs, Fable 5 and Mythos 5, marks a historic first, impacting not just foreign nationals but also US allies like the UK, Canada, Australia, and New Zealand.

• SkillSpector

  • SkillSpector is a security scanner for AI agent skills, revealing that 26.1% of skills harbor vulnerabilities and 5.2% exhibit malicious intent, thus ensuring safer installations.

• Building an Open Source Edge Semantic Cache for LLMs in Rust/WASM – Sanity check on the architecture?

  • The proposed architecture aims to create a lightweight, zero-dependency semantic cache at the CDN Edge using Rust/WASM, significantly reducing latency for real-time LLM workloads by avoiding centralized gateways and costly API calls.

• GLM 5.2 Is Out

  • GLM-5.2 features a 1-million-token context window, enhancing its ability to handle complex coding tasks and long-horizon projects, making it a significant upgrade for developers.

• NVIDIA Blackwell Leads on First Agentic AI Infrastructure Benchmark

  • NVIDIA Blackwell Ultra NVL72 outperforms competitors by running up to 20x more agents per megawatt than NVIDIA Hopper, showcasing its superior capability in handling complex agentic AI workloads.

• Derivative-Free Neural Network Optimization: MNIST Case