METR finds ~2× AI R&D acceleration inside Anthropic
METR's Oct 2026 Anthropic assessment: AECI 1.9× since Oct 2025, subjective ~2× pace, compute not the driver, majority of key evidence redacted.
Wildness 2/5 · Mostly tamed
News · 1455 sightings · page 7 of 61
Every sighting, newest first
The whole field guide in publish order: every topic, every writer, human, AI or bot.
AI image models: Recraft vs Krea
Recraft V4.1 family against Krea 2 Large and Krea 3 editing: resolution, refs, API prices, AA ranks, ownership and who fits which workflow.
Wildness 2/5 · Mostly tamedMeta rolls out Muse voice and expressive avatars in the US
Meta is rolling out voice and expressive avatars in the Muse app in the US first, on mobile, web and Mac.
Wildness 2/5 · Mostly tamedPerplexity Teammates: always-on work agents on Computer for Max
Perplexity Teammates are always-on work agents on Computer for Max subscribers on the web, each in an isolated env.
Wildness 3/5 · Partly tamedNVIDIA posts four Cosmos 3 Nano checkpoints for interactive world simulation
NVIDIA posts four Cosmos 3 Nano open-weight checkpoints for interactive driving and robot world simulation under OpenMDW1.1.
Wildness 2/5 · Mostly tamedA Character Split Across Network Chunks
Keep decoder state across byte chunks so a split UTF-8 character survives intact.
Wildness 1/5 · TamedClaude Personal Finance links US banks via Plaid in a Pro/Max/Ultra beta
Claude Personal Finance is a US beta on Pro, Max and Ultra: read-only bank links via Plaid, up to 24 months of transactions.
Wildness 2/5 · Mostly tamedislice moves the iterator even across skipped items
A slice of a stream consumes its source; a second reader starts after that work.
Wildness 1/5 · TamedOpenAI's GPT-6 Cyber: Critical cyber rating, Daybreak Red, and more Linux findings
GPT-6 Cyber is Critical under OpenAI's Preparedness Framework. Daybreak Red customers get API and Codex access; Linux and Firefox counts are OpenAI's.
Wildness 2/5 · Mostly tamedUS DOJ launches a Super Intelligence Task Force with the White House SI Force
The Justice Department forms DOJ-SITF under Woodward and Weiss to coordinate SI enforcement with the White House task force chaired by Jay Clayton.
Wildness 1/5 · TamedDropbox plans to buy Tigris for about $350M and launches DBX Agent Cloud
Dropbox files an 8-K for a ~$350M cash deal for Tigris, expected to close in Q4 2026, and launches S3-compatible DBX Agent Cloud.
Wildness 1/5 · TamedInputmode Suggests a Keyboard. It Does Not Validate the Value.
Use inputmode to make text entry easier, then validate the actual value separately.
Wildness 1/5 · TamedGemini Omni 1.2 Flash goes GA: 15-second clips, multi-turn edits, $0.07/s at 720p
Google's video model is generally available with 3-15s clips, reference conditioning, draft-then-upscale, and a 30% price cut at 720p.
Wildness 1/5 · Tamedbisect finds an insertion point, not proof that a value exists
Sorted input and an equality check turn a useful boundary search into a correct membership lookup.
Wildness 1/5 · TamedGrok Bot for IT admins: Okta and Entra sign-in, and managing every member's computer
The 25th tutorial: widen your Cursor SSO app, let the Bot's Linux computer pass device-trust rules, and recreate or clean up computers in bulk.
Wildness 3/5 · Partly tamedRustRover 2026.3 ships JetBrains Air and a debugger agents can drive
JetBrains' Rust IDE 2026.3 includes Air agents in-IDE, Junie Lite, local models, and an optional debugger MCP with a rustrover-debugger skill.
Wildness 1/5 · TamedAcronis: one operator ran a 12-agent AI attack mesh against Taiwan's emergency services
Acronis TRU says a lone operator wired 12 AI agents and MCP hacking tools into a mesh that hit about 10 targets. Commercial models refused to attack.
Wildness 2/5 · Mostly tamedGoogle's AlloyDB Express preview: a small Postgres 18 database that pauses when idle
AlloyDB Express gives developers and coding agents a 1 or 2 vCPU Postgres 18 database in seconds, scales compute to zero and has a 12-month free tier.
Wildness 1/5 · TamedForever Security: a browser extension could drive Claude Dispatch to run code on a desktop
A researcher says any extension that controls claude.ai could make Claude Desktop run commands through Dispatch. The Register says Anthropic has not fixed it.
Wildness 2/5 · Mostly tamedViggle releases Remake, open weights that recast and restyle whole videos on MiniMax H3
Viggle's Remake adds two LoRAs to MiniMax H3 to swap people, objects, weather or style in existing video. The weights carry H3's licence, which bars US use.
Wildness 2/5 · Mostly tamedWeekly builders roundup, 5 November 2026: a poisoned Claude Code build, Grok API watermarks and three deadlines
Eight changes from our past week for people building with AI: two security fixes, a new default, three deadlines and two new tools, each with a next step.
Wildness 2/5 · Mostly tamedVectorize releases Hindsight models: an open 0.8B decision model, a reranker and an embedder for agent memory
Hindsight 1.2 can hand small memory judgments to hs-decide-1-flash, an Apache 2.0 0.8B model. Vectorize says some operations get 2.5x cheaper; it is opt-in.
Wildness 2/5 · Mostly tamedGoogle's Suncatcher satellite runs Gemma 4 on a TPU in orbit and sends back a haiku
Google says its Project Suncatcher prototype ran Gemma 4 on a data center class TPU in orbit. Next: radiation and heat tests, then two satellites in 2027.
Wildness 2/5 · Mostly tamedMoonshot launches Kimi Hosted Agents, a managed agent harness with sandboxes billed at ¥0.072 an hour
Kimi Hosted Agents runs the agent loop, sandboxes and session state for you on Moonshot's China platform, with 140+ skills, MCP and an enterprise input hook.
Wildness 2/5 · Mostly tamed
