NO/FOMO

Independent AI signal, once a day

The AI briefing worth opening.

ISSUE DATE2026-06-13DEFAULT EDITION
This issue
—
All time
—

AI Blog

2 stories
01

Responds to US Government Directive Suspending Access to Fable 5 and Mythos 5

Anthropic issued an official response to a US government export control directive requiring the immediate suspension of all access to its Fable 5 and Mythos 5 models for foreign nationals and customers. The government cited national security concerns over potential safeguard bypasses, which Anthropic clarified are narrow, non-universal jailbreaks with capabilities already widely available in other models. This geopolitical policy shift treats nationality as a hard boundary for model access, as discussed by industry observers, highlighting Europe's technological dependence. Access to all other Anthropic models remains unaffected. (source: https://www.anthropic.com/news/fable-mythos-access)

Fable 5Mythos 5Export ControlModel SafeguardsJailbreakingLarge Language ModelTech NewsIndustry News
02

Ubiquitous Workplace Recording Emerges as a New System of Record

Andreessen Horowitz partner David Haber outlined how AI-driven workplace recording is transforming conversational audio into structured corporate context. Modern AI systems use these recordings to learn corporate culture and edge cases, enabling autonomous AI agents to stand in for executives and perform productive tasks across organizations. Companies such as OpenAI and Bridgewater Associates are currently adopting this methodology to build institutional context. This transition represents a new enterprise software category organized around voice rather than text, aiming to improve individual productivity and deliver top-down alignment oversight for executives. (source: https://a16z.com/everything-is-recorded-now/)

Workplace RecordingEnterprise AIAI AgentsVoice DataSystem of RecordArtificial IntelligenceGenerative AIIndustry News

Hacker News

8 stories
01

Statement on US government directive to suspend access to Fable 5 and Mythos 5

Anthropic has issued an official statement addressing a United States government directive to suspend access to its advanced AI models, Fable 5 and Mythos 5. This regulatory intervention marks a significant escalation in state-level oversight of frontier generative AI systems, directly impacting developers and enterprise users. The action has ignited widespread industry discussion regarding digital borders, model compliance, and national security policies in technology development. (source: https://www.anthropic.com/news/fable-mythos-access, discussion: https://news.ycombinator.com/item?id=48519780, discussion: https://news.ycombinator.com/item?id=48513536)

AnthropicFable 5Mythos 5AI GovernanceGovernment DirectiveArtificial IntelligenceLarge Language ModelGenerative AI
02

Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models

Discussions between Amazon's Chief Executive Officer and United States government officials reportedly triggered a regulatory and security crackdown on Anthropic's Claude series of artificial intelligence models. The talks focused on potential vulnerabilities, intellectual property sovereignty, and export control policies governing highly capable large language models. This intervention underscores the increasing geopolitical scrutiny and tension between frontier AI labs, cloud hosting providers, and national security agencies. (source: https://www.wsj.com/tech/ai/amazon-ceos-talks-with-u-s-officials-triggered-crackdown-on-anthropic-models-dcc90578?st=Yct6gx&reflink=desktopwebshare_permalink)

AnthropicAmazon Web ServicesAI RegulationNational SecurityCloud ComputingArtificial IntelligenceGenerative AILarge Language Model
03

GLM 5.2 Is Out

The research community has officially released GLM 5.2, the latest iteration in the General Language Model series of open-source bilingual conversational models. This update delivers targeted optimizations in inference speed, contextual understanding, and multi-turn dialogue capabilities for both English and Chinese text processing. Developed as a competitor to industry-leading foundation models, it features structural adjustments that reduce computational latency while maintaining high precision across reasoning tasks. (source: https://digg.com/tech/ii9xibgn)

GLM 5.2General Language ModelBilingual AIModel InferenceOpen-Source LLMArtificial IntelligenceLarge Language ModelNatural Language Processing
04

RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8

An independent developer configured an NVIDIA RTX 5080 and RTX 3090 dual-GPU setup to run the Qwen 3.6 27B large language model. By using 8-bit quantization, the consumer-grade system achieved local inference speeds of 80 tokens per second. This demonstrates the viability of utilizing heterogeneous, consumer-budget hardware to run highly capable, quantized open-source models at high throughput without relying on enterprise cloud resources. (source: https://imil.net/blog/posts/2026/rtx-5080-+-rtx-3090-setup-80+-tok-s-on-qwen-3.6-27b-q8/)

RTX 5080RTX 3090Qwen 3.6 27BModel QuantizationLocal InferenceArtificial IntelligenceLarge Language ModelGenerative AI
05

Shepherd's Dog: A Game by Fable

Developer Koen van Gilst built and launched 'Shepherd's Dog', an interactive web-based game created using Fable, a toolset that integrates with Anthropic's Claude AI assistant. The project demonstrates the viability of using generative AI and autonomous agents for rapid game prototyping. By interpreting natural language prompts to orchestrate complex game logic, physics, and rendering directly in a web browser, the tool showcases a shifting workflow in digital media creation. (source: https://koenvangilst.nl/lab/claude-fable-shepherds-dog)

Fable AIClaude AIGame PrototypingGenerative AIWeb DevelopmentArtificial IntelligenceGenerative AIAI Agent
06

Show HN: Paca – Lightweight Jira alternative for human-AI collaboration

An open-source developer released Paca, a lightweight, Go-based alternative to Jira designed for human-AI collaboration in agile projects. The platform treats humans and AI agents as equal teammates, allowing for joint sprint planning and task assignment. Built with custom views and a WebAssembly (WASM)-based plugin architecture, the system enables secure, modular extensions in multiple languages to integrate agentic workflows directly into project management. (source: https://github.com/Paca-AI/paca)

Project ManagementGo LanguageWASMAI AgentsAgile DevelopmentArtificial IntelligenceAI AgentOthers
07

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

The developer community observed that TensorZero, an open-source AI development platform, suddenly archived its primary GitHub repository. This transition occurred immediately after the project raised a $7.3 million seed funding round. The rapid change in project governance from active open source to archived status has raised significant concerns among developers about the commercialization pressures and long-term viability of venture-backed open-source software in artificial intelligence. (source: https://github.com/tensorzero/tensorzero)

Open Source SoftwareVenture CapitalDeveloper ToolsAI PlatformRepository GovernanceArtificial IntelligenceLarge Language ModelOthers
08

AI coding at home without going broke

Developer Stephen Bochinski has outlined cost-effective workflows for running AI-assisted coding setups at home without expensive subscriptions. The guide details how to leverage smaller, open-source local large language models on consumer hardware, combined with pay-as-you-go APIs for complex software development tasks. By optimizing local IDE integrations and model selection, developers can reduce recurring costs while maintaining a highly responsive, custom programming environment. (source: https://stephen.bochinski.dev/blog/2026/06/13/ai-coding-at-home-without-going-broke/)

AI codingLocal LLMCost optimizationOpen-source modelsAPI integrationArtificial IntelligenceLarge Language ModelGenerative AI

Twitter

6 stories
01

US Government Imposes Export Controls on Anthropic AI Models

The United States government has issued a formal export control directive targeting Anthropic's flagship large language models, Fable 5 and Mythos 5, citing national security concerns. The directive prohibits access to these models by foreign nationals globally, including those employed by Anthropic, forcing the company to suspend general access to the Claude Fable 5 platform. New user sessions will default to the Opus 4.8 framework. This regulatory compliance action represents a significant shift in government oversight and operational constraints for frontier AI laboratories. (source: https://x.com/AnthropicAI/status/2065597531644743999)

AnthropicExport ControlNational SecurityAI RegulationComplianceArtificial IntelligenceLarge Language ModelIndustry News
02

Cohere Releases New Lightweight 30B Open-Weight Model For Agentic Coding

Cohere has released a new lightweight, open-weight model specialized for agentic coding. The model is built on the Command A+ architecture and features a 30-billion parameter footprint alongside a parallel transformer design. Although it has roughly half the parameter count of larger variants, the model incorporates a higher density of layers to optimize performance for complex developer workflows and autonomous programming assistants. (source: https://x.com/rasbt/status/2065778965273354545)

CohereOpen-Weight ModelAgentic CodingParallel TransformerCommand A+Artificial IntelligenceLarge Language ModelOpen Source
03

Google DeepMind Launches New Robotics Accelerator With Fifteen Startups

Google DeepMind has launched a new Robotics Accelerator program in Europe to support fifteen selected startups working on physical AI and embodied systems. The accelerator will provide resources and collaborative support to integrate advanced machine learning models into robotics software, aiming to solve real-world industrial challenges and advance European hardware innovation. (source: https://x.com/ZoubinGhahrama1/status/2065585793108722146)

Google DeepMindRoboticsPhysical AIAccelerator ProgramStartupsRoboticsArtificial IntelligenceIndustry News
04

OpenAI Faces Formal Investigation by State Attorneys General

A coalition of state attorneys general has launched a formal joint investigation into OpenAI's practices, safety, and operational accountability. The probe escalates legal and government oversight on the company's generative AI systems. This regulatory action reflects a rising trend of state-level investigations targeting the deployment and risk management of foundational large language models. (source: https://x.com/GaryMarcus/status/2065677831787721075)

OpenAIInvestigationLegalRegulationState Attorneys GeneralArtificial IntelligenceGenerative AIIndustry News
05

Kling AI Showcases Surprising Character Animation Capabilities

Kling AI has released a video demonstration of its generative video synthesis model, highlighting advancements in character animation and physics consistency. The demonstration highlights the model's capacity to render natural expressions and maintain structural coherence across dynamic frame sequences, aiming to improve outputs for digital media and cinematic productions. (source: https://x.com/Kling_ai/status/2065811435892924604)

Kling AIGenerative VideoCharacter AnimationAI Video GenerationMotion SynthesisArtificial IntelligenceGenerative AIVideo Understanding
06

DeepMind Seeks Research Engineer for Multi-Agent Learning Initiatives

Google DeepMind has opened recruitment for a Research Engineer to join its multi-agent learning division. The engineer will focus on developing complex coordination, social learning, and strategic reasoning systems inside multi-agent reinforcement learning architectures to improve cooperation and decision-making among autonomous agents in shared virtual environments. (source: https://x.com/ZoubinGhahrama1/status/2065884467550400662)

DeepMindResearch EngineerMulti-agent LearningReinforcement LearningHiringArtificial IntelligenceMachine LearningAI Agent