Eclecta — daily brief

Daily brief · Tuesday, June 23, 2026

Tue, 23 Jun 2026 00:00:00 GMT

A new theory recasts prompt injection as role confusion the model can be tricked out of; the top open-weight model ships text-only; and a Codex logging default writes terabytes to local SSDs.

Daily brief · 2026-06-23 · Read on Eclecta

Daily brief · Monday, June 15, 2026

Mon, 15 Jun 2026 00:00:00 GMT

The US pulls two Anthropic models worldwide over a disputed code-review jailbreak, the first export-control takedown of a deployed model, as AI-found zero-days pile up in FFmpeg and the Pixel 9.

Daily brief · 2026-06-15 · Read on Eclecta

Daily brief · Saturday, June 13, 2026

Sat, 13 Jun 2026 00:00:00 GMT

Anthropic reverses a covert policy that silently degraded Claude Fable's answers for suspected AI researchers, as new work undercuts both multi-agent systems and the probes meant to catch models lying.

Daily brief · 2026-06-13 · Read on Eclecta

Daily brief · Friday, June 12, 2026

Fri, 12 Jun 2026 00:00:00 GMT

Project Zero prices a full Pixel root chain at roughly eleven person-weeks and documents months of patch lag, as fresh benchmarks measure how far AI agents still fall short on real work.

Daily brief · 2026-06-12 · Read on Eclecta

Daily brief · Thursday, June 11, 2026

Thu, 11 Jun 2026 00:00:00 GMT

A German court strips AI summaries of search's legal shield; Anthropic ships its most capable model behind heavy filters while its CEO asks to be regulated; and new research shows alignment passing benchmarks it quietly fails underneath.

Daily brief · 2026-06-11 · Read on Eclecta

Daily brief · Wednesday, June 10, 2026

Wed, 10 Jun 2026 00:00:00 GMT

Anthropic ships a frontier model that reroutes dual-use queries instead of refusing them, Amazon deploys random-graph datacenter networks at scale, and error messages emerge as a privileged prompt-injection surface.

Daily brief · 2026-06-10 · Read on Eclecta

Daily brief · Monday, June 8, 2026

Mon, 08 Jun 2026 00:00:00 GMT

A researcher reads two decades of encrypted military traffic hidden in the public GPS signal, OpenAI and Simon Willison both move to contain untrusted input to LLMs, and a $280 soundbar becomes a remote keyboard.

Daily brief · 2026-06-08 · Read on Eclecta

Daily brief · Friday, June 5, 2026

Fri, 05 Jun 2026 00:00:00 GMT

Hugging Face rebuilds its CLI for coding agents and benchmarks the token cost of hand-rolled alternatives; a preprint caps eval scores to expose agents that game the test; NVIDIA releases an open multimodal guardrail.

Daily brief · 2026-06-05 · Read on Eclecta

Daily brief · Thursday, June 4, 2026

Thu, 04 Jun 2026 00:00:00 GMT

Cloudflare finds about half of Tier 1 networks accept forged BGP paths; Microsoft fields a from-scratch model family at Build; Uber caps coding agents at $1,500 a month.

Daily brief · 2026-06-04 · Read on Eclecta

Daily brief · Wednesday, June 3, 2026

Wed, 03 Jun 2026 00:00:00 GMT

Microsoft announces a seven-model MAI family backed by a rare, transparent training report; Alphabet raises about $80 billion, including Berkshire's first big Google stake, to fund the compute race.

Daily brief · 2026-06-03 · Read on Eclecta

Daily brief · Tuesday, June 2, 2026

Tue, 02 Jun 2026 00:00:00 GMT

An interpretability preprint says diffusion image models read only word meaning and order from prompts, a Lean4 framework brings formal verification to agent workflows, and attackers seized Instagram accounts by asking Meta's support bot.

Daily brief · 2026-06-02 · Read on Eclecta

Daily brief · Monday, June 1, 2026

Mon, 01 Jun 2026 00:00:00 GMT

Two frontier labs detail how they measure and contain their agents; a Zapier exploit chain and Vercel's "inference theft" show what weak containment costs; and reverse-engineers read microcode and hidden memory off the silicon.

Daily brief · 2026-06-01 · Read on Eclecta

Daily brief · Friday, May 29, 2026

Fri, 29 May 2026 00:00:00 GMT

Anthropic's $65 billion raise and an incremental Opus 4.8 lead a quiet day, with new research showing coding agents leaking secrets and firing real attacks at live sites.

Daily brief · 2026-05-29 · Read on Eclecta

Daily brief · Thursday, May 28, 2026

Thu, 28 May 2026 00:00:00 GMT

OpenCode's founder picks apart the pitch that AI lifts team output, Stratechery sizes up satellites as server racks, and Cisco Talos open-sources synthetic security logs that stay consistent across 20-plus formats.

Daily brief · 2026-05-28 · Read on Eclecta

Daily brief · Tuesday, May 26, 2026

Tue, 26 May 2026 00:00:00 GMT

Huawei pitches an architecture-first scaling law to skirt EUV denial, the memory supercycle prices sub-$100 phones out of emerging markets, and Google's AI search box draws a reported migration to rivals.

Daily brief · 2026-05-26 · Read on Eclecta

Daily brief · Monday, May 25, 2026

Mon, 25 May 2026 00:00:00 GMT

A maintainer puts hard numbers to open source's agent-traffic problem, an AI disproves an 80-year-old Erdős conjecture, SPEC's new CPU benchmark gets its first independent teardown, and a CISA contractor publishes the agency's own cloud keys.

Daily brief · 2026-05-25 · Read on Eclecta

Daily brief · Friday, May 22, 2026

Fri, 22 May 2026 00:00:00 GMT

Microsoft Research releases a codesigned small-model agent stack and claims it leads computer-use benchmarks it ran itself.

Daily brief · 2026-05-22 · Read on Eclecta

Daily brief · Thursday, May 21, 2026

Thu, 21 May 2026 00:00:00 GMT

An OpenAI reasoning model produces an externally verified disproof of a 1946 Erdős conjecture; a GitHub employee's poisoned IDE extension exposes about 3,800 internal repos; and an essay rereads China's AI optimism as fear of falling behind.

Daily brief · 2026-05-21 · Read on Eclecta

Daily brief · Wednesday, May 20, 2026

Wed, 20 May 2026 00:00:00 GMT

Google sends its agentic science assistant to Nature and into Gemini for Science, Anthropic splits agent brains from hands on Cloudflare, and a new lattice-QCD result quietly closes the muon g−2 anomaly.

Daily brief · 2026-05-20 · Read on Eclecta

Daily brief · Tuesday, May 19, 2026

Tue, 19 May 2026 00:00:00 GMT

Two vendor field reports put a security-tuned Anthropic model preview to work on real codebases and credit the scaffolding over the model, as Marc Brooker reframes where coding agents win.

Daily brief · 2026-05-19 · Read on Eclecta

Daily brief · Monday, May 18, 2026

Mon, 18 May 2026 00:00:00 GMT

Model internals own a quiet day: how 2026's open-weight LLMs cut long-context cost, two sober takes on RL and steering, and Gemini 3.5 Flash ships.

Daily brief · 2026-05-18 · Read on Eclecta

Daily brief · Friday, May 15, 2026

Fri, 15 May 2026 00:00:00 GMT

A hidden lock in ClickHouse query planning stalled Cloudflare's billing; AI turns up on both sides of the CVE curve; and new open releases put their gains down to better data, not bigger models.

Daily brief · 2026-05-15 · Read on Eclecta

Daily brief · Thursday, May 14, 2026

Thu, 14 May 2026 00:00:00 GMT

OpenAI hand-builds a Windows sandbox for its Codex agent and discloses an npm worm that forced a code-signing certificate rotation, while Microsoft Research opens up the mimalloc allocator.

Daily brief · 2026-05-14 · Read on Eclecta