imo bluesky has a pretty good take on AI in society, generally, and i’m excited to see where our conversation develops
Information Center
Yandex (Russia's Google) released Yambda-5B dataset so the researchers can develop and eval recommendation models at scale. It is one of the world's largest open datasets of its kind, it contains 4.79 billion anonymized interactions from 1 million users across 9.39 million tracks from Yandex Music.
This study finds that Moonshot AI's Kimi K3.1 Pro is bench maxxed (Terminal Bench tasks). Study: r40.org/papers/kimi_... (arxiv version shortly) Data: huggingface.co/datasets/R40...
Tencent released Hy4 (Open weights coming soon) ◆ First Hy model with eyes: native vision + computer use. ◆ The price stayed in preview. The benchmarks didn't. ◆ Open weights coming soon. Hy4 Flash right behind. Blog: hy.tencent.ai/research/hy4...
Baidu's (China's Google) released ERNIE 5.5 Flash Base as open-weight A 280B-parameter model with 14B active per token. Its design includes: • Latent-Gated Sparse Attention • Multi-Router Latent MoE • ECHO • Model: huggingface.co/baidu/ERNIE-... • Tech blog: ernie.baidu.com/zh/release/e...
Forget legs and wheels. Neufangled Robotics hung its home robot from the ceiling. Stringman is an open source claw on four cables: $795, lifts 2 kg, covers a whole room, and learns tidying tasks with LeRobot. Source: The Register Video: Neufangled Robotics #Robotics #OpenSource #HomeRobotics
Anne Hathaway, Madonna and others have backed The New York Times in its legal fight against OpenAI and Microsoft, with a coalition of entertainment industry groups filing a brief arguing that AI firms shouldn't use publishing content to train models www.axios.com/2026/11/03/h...
Puzzles and games can reveal gaps in models’ smarts. Try your hand at these seven tests to see if you can outwit AI.
In various tests, solo AI agents invented versions of basketball and “the floor is lava.” But were they truly playing?
Monitoring apps promise to keep young people safer online, but looking in on kids’ phones can backfire. Online safety experts say there’s a better way.
The open-closed AI debate is more complicated than it seems.
In the aftermath of a cyberattack carried out by a swarm of OpenAI agents, the heads of top labs agree they need to slow down. But they won't open up.
What do last night’s resounding midterm wins for Democrats tell us about the future of AI? Here are some of my initial takeaways: 1. AI was a key issue for voters this year - with 56% saying in exit polls that it was crucial or a major factor for them.
They find that when you put a heart ❤️ next to a coding problem, and Claude & GPT write recursive code 100% of the time (0% baseline). A Heart Means Recursion
OpenAI is discovering that there's money to be made by playing both sides of the cybersecurity game. - Sell AI tools that can be used to hack websites and exploit software vulnerabilities. - Sell AI tools that protect those same websites and software from cyberattacks and vulnerabilities.
OpenAI just poached the product lead for Claude Code. I was wondering why a Claude Code product leader would leave pre-IPO Anthropic shares on the table. The answer was pre-IPO OpenAI shares.
$38M says the next logistics fleet has no crew and no surface. Corridian just left stealth to build it. Its ultra-large robot submarines carry container payloads thousands of nautical miles underwater. Point72 led the seed. Source: Business Wire Image: Corridian #Robotics #DefenseTech #Autonomy
LOL. Cat Wu (a former Claude Code product manager) is joining OpenAI.
a few of these are still missing
ChatGPT Images 2.6 can do GIFs openai.com/index/introd...
Anthropic reminds you that there’s other agent harnesses that don’t suck
Some of the largest AI infrastructure projects in the world are underway in Texas. A dramatic political shift could slow their construction—and deepen concerns about the economics of the AI boom.
Newly relevant: Last month I went to see Musk's new data center complex in Mississippi myself. You cannot understand just how far ahead he is of other neocloud builders until you see this kind of thing up close. SpaceX is making $2 billion+ a month on this... www.youtube.com/shorts/7aZQt...
Aformic just raised $6.95M to move stuff inside factories. The Duncan, SC startup's seed, led by Veteran Ventures, is no pre-revenue bet: $10M in revenue last year, robots already running at Tier 1 defense makers. Source: Tectonic Defense Image: Aformic #Robotics #Manufacturing #DefenseTech
i struggle to write about cybernetics i have a blog post about Ashby's Law that i've tried writing like 5 times, just tried again yesterday and i'm probably trashing it again this time too
AI semiotics: cool paper suggests that AIs are strongly impacted by visual symbols in ways they do not always recognize in their own chains-of-thought. The way they react to those symbols is very different than how a human might interpret them as well. arxiv.org/pdf/2610.364...
Boston Dynamics rebuilt Stretch around one metric: uptime. Gen 2 was redesigned from 100,000+ operating hours: 81% less cabling, up to 1,000 cases an hour, and a drive swap that now takes an hour, not a day. Source: Boston Dynamics Image: Boston Dynamics #Robotics #WarehouseAutomation #Logistics
Blue Texas is kerrrrrazy
The home robot that hides in plain sight as a lamp. SYNCERE just raised $11M pre-seed led by a16z Speedrun. LUME folds laundry, tidies and makes the bed, starting at $2,699. It is already in homes and learning from them. Source: Business Wire Image: SYNCERE #Robotics #HomeRobotics #PhysicalAI
Needle for Andrews in South Carolina now up to 93 percent
Private to this visit. Text and context are processed by AI.