today's signal · no scroll
the brief
Policy shock dominated the day as the US ordered Anthropic to suspend access to Fable 5 and Mythos 5 for all foreign nationals, a stark precedent for model governance. Meanwhile, the stack kept moving: AllenAI/Hugging Face shipped olmo-eval, Vercel embedded Workflows in Nitro v3, Cloudflare detailed a 10x scale-up, and practical hacks—from hnsw in SQLite to Tailscale on e-readers—advanced real-world DX.
Citing national security authorities, the US ordered Anthropic to suspend all access to Fable 5 and Mythos 5 by any foreign national, forcing immediate customer impact and compliance reviews.
US Government directive to suspend access to Fable 5 and Mythos 5 — Article URL: https://www.anthropic.com/news/fable-mythos-access Comments URL: https://news.ycombinator.com/item?id=48511072 Points: 1033 # Comments: 634
Willison highlights the unprecedented breadth of the order—blocking named models for all foreign nationals—and its chilling signal for model governance and developer dependencies.
Statement on the US government directive to suspend access to Fable 5 and Mythos 5 — <p><strong><a href="https://www.anthropic.com/news/fable-mythos-access">Statement on the US government directive to suspend access to Fable 5 and Mythos 5</a></strong></p> Well this is <em>nuts</em>:</p> <blockquote> <p>The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside...
AllenAI and Hugging Face released olmo-eval, a reproducible evaluation workbench that plugs into the model development loop for standardized runs, comparisons, and iteration across datasets and tasks.
olmo-eval: An evaluation workbench for the model development loop
Workflow SDK steps now run in the same Nitro v3 runtime as your app with direct server API access, plus a built-in /_workflow UI for inspecting and debugging runs.
Workflow SDK now runs natively in Nitro v3 — 's native Nitro v3 integration is now in beta. Steps run inside the same bundled runtime as the rest of your app, instead of a separate bundle. Nitro's and other server-side APIs work directly inside functions.Workflow SDKuseStorage()"use step" The Nitro dev server also serves the workflow web UI at . Open it in your browser to inspect, monitor, and debug workflow runs./_workflow Workflow routes are now bundled by Nitro as part of the app build. De...
Bytedance’s Deer-Flow offers an OSS long-horizon agent harness with sandboxes, memories, tools, skills, subagents, and a message gateway to coordinate tasks that span minutes to hours.
GitHub - bytedance/deer-flow: An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours. — An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours. - bytedance/deer-flow
A SQLite extension that embeds HNSW vector indexing enables fast semantic retrieval locally—ideal for small apps, desktop tools, and edge deployments without external vector databases.
hnsw-sqlite — A quietly trending SQLite extension that adds HNSW vector search directly inside SQLite, making it relevant for AI retrieval and local developer tooling.
Adds language-aware session titles, configurable regex link badges, better Bedrock credential caching, and stricter model alias enforcement—useful quality-of-life fixes for teams managing Claude Code.
v2.1.176 — What's changed Session titles are now generated in the language of your conversation (set the language setting to pin a specific language) Added footerLinksRegexes setting for regex-matched link badges in the footer row, configurable via user or managed settings Improved Bedrock credential caching: credentials from awsCredentialExport are now cached until their Expiration instead of a fixed 1 hour Fixed availableModels enforcement: alias model picks can no longer be redirected to a...
You can now run Tailscale proxies and SSH on jailbroken Kindles, with a plugin extending support to Kobo and Pocket readers—handy niche endpoints for private networks.
More Tailscale tricks for your jailbroken Kindle — Tailscale on jailbroken Kindles now supports proxies and SSH. And a plugin supports Kobos and Pocket Readers, too.
Engineering write-up on scaling Security Insights from 12 to 120 scans/sec by optimizing Kafka consumers, Postgres queries, and APIs—hitting a 10x throughput increase without new hardware.
Scaling Security Insights: how we achieved a 10x increase in global scanning capacity — Cloudflare Security Insights system now processes over 120 scans per second, providing frequent insights for all customers. By optimizing Kafka consumers, Postgres queries, and our API, we scaled our throughput 10x without adding hardware.
An updated demo shows OpenAI’s realtime audio sessions grounded with document context from files and URLs, enabling more useful voice interactions for retrieval and agentic flows.
OpenAI WebRTC Audio Session, now with document context — <p><strong><a href="https://tools.simonwillison.net/openai-webrtc">OpenAI WebRTC Audio Session, now with document context</a></strong></p> I built the first version of this tool <a href="https://simonwillison.net/2024/Dec/17/openai-webrtc/">in December 2024</a> to try out the then-new OpenAI WebRTC API for interacting with their realtime audio models.</p> <p>Last month OpenAI <a href="https://openai.com/index/advancing-voice-intelligenc...
