Morning brief

Saturday, June 13

2 people · 0 companies · 3 topics

TL;DR The US Department of Commerce forced Anthropic to suspend Fable 5 and Mythos 5 globally on June 12 — the first government-mandated AI model recall on record. Moonshot AI also dropped Kimi K2.7-Code overnight, an open-weight 1T-param coding model with contested self-reported benchmarks.

Models & research

  • The US DoC issued an export control directive at 5:21 PM ET on June 12, ordering Anthropic to suspend all access to model-releases Claude Fable 5 and Mythos 5 for any foreign national. Because Anthropic cannot separate foreign nationals in real time, it pulled both models for everyone worldwide. The directive cited a narrow jailbreak — asking the model to review a codebase and flag software vulnerabilities. Anthropic complied but pushed back publicly: it disputes that a narrow, non-universal jailbreak justifies a global recall, and warns that applying this standard industry-wide “would essentially halt all new model deployments.” The shutdown came three days after launch and 11 days after Anthropic filed confidentially for an IPO; pre-IPO contracts on Hyperliquid dropped ~3.7% to ~$1,627. Anthropic statement · TechCrunch · CNBC · Fortune · HN

  • Moonshot AI open-sourced Kimi K2.7-Code on June 12: 1T-parameter MoE (32B active, 384 experts), 256K context, Modified MIT license, API at $0.95/$4.00 per M tokens. Self-reported gains are +21.8% on Kimi Code Bench v2, +31.5% on MLS Bench Lite, 30% fewer reasoning tokens vs K2.6. All three benchmarks are proprietary; no independent SWE-bench or LiveCodeBench numbers exist yet, and practitioners are calling that out. VentureBeat · MarkTechPost

People

  • simonw documented the Fable 5 cutoff to the minute: the model was live at 6:58 PM Pacific, returned a 404 at 6:59 PM. He called the government action “absurd” in his June 13 writeup. simonwillison.net

  • natolambert posted Substack reflections on “the current state of the Claude Fable saga,” noting a cluster of overlapping problems on different timelines. His earlier criticism of Anthropic’s silent model manipulation (June 11 Substack) fed into the backdrop of the week’s chaos. Substack note

OSS

  • GitHub trending (Python, today): anthropics/skills leads at 150K total stars, +459 today. NVIDIA/SkillSpector — a security scanner for agent skills — is surging fast at +809 stars, 4K total. browser-use still climbing at +181, 98K total. GitHub Trending

Other

  • Polymarket: GPT-5.6 release before June 15 sits at 21% — traders moved the odds down sharply this week. The “released by July 31” market is at 96% on $904K volume. OpenAI has not made an official announcement. (Polymarket: 96% yes by July 31, $904K volume) Polymarket · KuCoin