Morning Brief: Sunday, June 14

Two-week window across 49 tracked feeds, scored against active research threads. Metadata only: titles, links, dates. Read the source for substance. (what we track, how we crawl, subscribe)

The Fable arc enters its geopolitical phase. The US suspension order (day 5) has triggered second-order effects: India is debating whether to build sovereign AI capacity or accept dependency on US-controlled model access; Meta is reportedly unwinding its $2B Manus deal after Beijing demanded terms incompatible with the new US restrictions. Separately, Anthropic publishes research on making Claude a chemist, domain-specific capability work that continues regardless of the Fable/Mythos access situation. UK police are accused of using AI to fabricate evidence, a concrete case of the governance failure mode the Fable debate has been theorising about. Weekend: no arXiv papers.

Top (5-7 min)

Meta reportedly moves to unwind $2B Manus deal after Beijing's demand: TechCrunch, 2026-06-14. The Fable suspension's second-order effects reach M&A: Beijing reportedly demanded terms that conflict with new US model-access restrictions, forcing Meta to unwind the deal. Geopolitical bifurcation of the AI supply chain accelerates.
As Anthropic suspends access to new models, India debates its AI future: TechCrunch, 2026-06-14. India as case study for the access-revocation problem: open source AI must win argued this yesterday in the abstract; India is now debating it concretely. Sovereign AI capacity vs. dependency on US API access.
Making Claude a Chemist: Anthropic via Hacker News, 2026-06-14. Anthropic's domain-specific capability research continues through the Fable crisis. Chemistry as a test domain for structured reasoning, a different safety question than the one dominating the news cycle.
UK Police Officer Accused of Using AI to Fake Evidence: Slashdot, 2026-06-14. Not hypothetical: a police officer allegedly used AI to fabricate evidence. This is the governance failure mode that the Fable debate has been theorising about: AI misuse by state actors. The UK case is about a rogue individual, not policy.
Don't trust large context windows: Hacker News, 2026-06-14. Practical limits of large context windows. Relevant to anyone building retrieval or agent systems that assume long-context fidelity.

Themes this week

Fable/Mythos geopolitical fallout (day 6): The arc: launch (06-09), invisible guardrails (06-10), Anthropic apologizes (06-11), proactivity bias (06-12), government suspension (06-13), geopolitical fallout (06-14): India debates, Meta/Manus unwinds. The arc has moved from product issues → safety concerns → regulatory action → international trade consequences.
AI governance: theory meets practice: UK police AI evidence fabrication is the concrete case the Fable debate has been theorising about. The SFT drives Gemini's safety properties post from Alignment Forum (06-13) provides technical context on how safety training actually works under the hood.
Developer tools weekend: Phoenix LiveView 1.2 released, significant for the Elixir real-time web ecosystem. Weave proposes merging based on language structure rather than lines. Relevant to anyone frustrated with merge conflicts in refactored code.

Scan (15 min)

Fable/Mythos arc (continuing)
- Meta moves to unwind $2B Manus deal, TechCrunch, 06-14
- India debates its AI future, TechCrunch, 06-14
- Making Claude a Chemist, Anthropic, 06-14
AI governance and safety
- UK police officer accused of using AI to fake evidence, Slashdot, 06-14
- SFT Drives Gemini's Safety Properties, Alignment Forum, 06-13
Developer tools and languages
- Phoenix LiveView 1.2 Released, Hacker News, 06-14
- Weave: Merging based on language structure, Hacker News, 06-14
- Software Architecture Guide, Hacker News, 06-14
- Free SQL→ER diagram tool, Hacker News, 06-14
- Four LTS Java versions get end-of-support 2029-2032, Slashdot, 06-14
Context windows and AI evaluation
- Don't trust large context windows, Hacker News, 06-14
Systems and hardware
- Double the VRAM of an RTX 3070, Hackaday, 06-14
- Robot Chess but Each Piece is a Small Robot, Hackaday, 06-14
- The Pacemaker Patch, Hackaday, 06-14
- Building a serial and VGA "everything console", Hacker News, 06-14
- Tribblix: The retro Illumos distribution, Hacker News, 06-14

Tail

Pac-Man, but you're the ghost, Hacker News, 06-14
Honda Civics and the Evil Valet, Hacker News, 06-14
How Dave Eggers avoids smartphones and Flock cameras, Slashdot, 06-14
Shareholder supremacy and the precog CEO, Pluralistic, 06-13
Increased Blob store limit for Hobby users, Vercel, 06-15

Feed silences (diagnostic)

arxiv-cs-ai: last items 06-12 (3258 in window), weekend. No papers expected until Monday.
anthropic-generated: last item 06-03 (Services Track, Partner Hub).
claude-code-releases: v2.1.177 (06-13), no new release today.
Apple ML Research: last item 06-08.
Terence Tao: 2 items in window (06-08, 06-09).
deepmind-blog: 6 items in window, last 06-11.
Ink & Switch: 1 item in window.
Jane Street: 1 item in window.

Build provenance

build: 2026-06-14 | crawler-sha: 508e4ab (Walsh-Research/1.2, compliance v1.3) | feeds: 49 core | items-considered: 4672 (14d, incl. 3258 arXiv) | warehouse: 14290 items | published: 25 | note: Fable arc day 6 geopolitical fallout; Meta/Manus $2B unwind; India sovereign AI debate; UK police AI evidence fabrication; weekend arXiv silence