@joho_no_todai Hacker jailbreaks Claude Fable like is a dollhouse. US government not impressed shuts it down to avoid national cybersecurity crisis: https://t.co/d4ZU3i1s50

@MaxForAI @kradleai Leaked: https://t.co/d4ZU3i1s50

@hmkz_ Hacker jailbreaks Claude Fable like is a dollhouse. US government not impressed shuts it down to avoid national cybersecurity crisis: https://t.co/d4ZU3i1s50

@PeteHegseth @DeptofWar @AnthropicAI Hacker jailbreaks Claude Fable like is a dollhouse. US government not impressed shuts it down to avoid national cybersecurity crisis: https://t.co/d4ZU3i1s50

Build Small hackathon closes in just 6 hours! We’ll disable updates & stop accepting new Spaces in the Hackathon Org after the deadline. Please do a final README check for tags, demo video, social link, & team member HF usernames. 👉Validate your app: https://t.co/DdUu6Ltw0m https://t.co/hkb6Xyj0K6

Americans realizing they spent $75 billion fighting Iran, then another $300 billion rebuilding Iran, just to reopen the Strait of Hormuz that was already open before the war started https://t.co/xEpusW60hX

Elon Musk’s est. lifetime philanthropic contributions range between $400M and $1B, or roughly 0.04% to 0.1% of his ~$1.1 trillion net worth. He doesn't retain it for his family, obviously. He keeps it so he can call himself a trillionaire. Compared to Gilded Agers? > https://t.co/Al14T5Qgut

We're launching a new research competition! Train the best pathology foundation model in just 1 hour! The competition has already been ongoing in @MedARC_AI community for a few weeks now and we've already been making a lot of great progress! Learn how to participate in the thread below...

@guansi True story: https://t.co/d4ZU3i1s50

World-Action Models (WAMs) have become the second dominant recipe for robot foundation models, next to classical VLAs. So where do they come from, and how do they compare vs VLAs? I wrote an small overview of the WAM landscape, with some personal takes: https://t.co/6S4gH9tWTt

BREAKING: 🇺🇸🇮🇷 JD Vance confirms that Iran will get $300 billion for war damages https://t.co/gWfCGf7AqJ

BREAKING: 🇺🇸🇮🇷 JD Vance confirms that Iran will get $300 billion for war damages https://t.co/gWfCGf7AqJ

LEARN TO CODE, THEY SAID Recent computer engineering grads now have the #2 highest unemployment rate of any major. Computer science is #4. Both sit above art history and English. So much for the safe degree.

Here’s a trend you can’t afford to ignore: the people who actually want AI to replace humanity. These “AI successionists” believe that artificial intelligence is our rightful heir — the next step in cosmic evolution. And they're more influential than you might think. 🧵1/n https://t.co/5GDux5nMLB

Every equity bubble inevitably becomes a credit bubble before it bursts... https://t.co/qkwJgOO9Tc

Every equity bubble inevitably becomes a credit bubble before it bursts... https://t.co/qkwJgOO9Tc

Today, we're excited to announce our first research competition! INTRODUCING NANOPATH: a framework and challenge to train the best pathology foundation model in just 1 hour! A quick thread on why we made this challenge and how to participate! https://t.co/hos10B7gFc

Built https://t.co/itho6xB6IN at #WeaveHacks4 🚀 A multi-agent system that analyzes logs, metrics, and traces, performs root cause analysis, learns from past incidents, and automatically creates GitHub PRs with proposed fixes. @OpenAI @cursor_ai @CopilotKit @wandb @Redisinc https://t.co/bgHcdc1Fqi

+2 more

@GaryMarcus If a government wants power instead of serving people, AI is the perfect Orwellian tool, fast, silent, and built for control. The Anthropic squeeze showed how easy it is to bend companies in the dark. This is exactly the kind of future Orwell warned about. https://t.co/RnzV7lfZns

.@iamtimdavis built a retro platformer where the terrain is generated by SIMD kernel computation. Meet GPU Boost Adventure. A neon, CRT-flavored endless runner. https://t.co/KFC9yR5oVw https://t.co/P99Fv4NbIW

Play it now at https://t.co/KFC9yR5oVw. Explore how it's made at https://t.co/P2TS9cUYgp. Reply with your high score!

Learning is not just about getting the right answer. Some of life’s most important lessons come from trial and error, frustration, collaboration and overcoming setbacks—experiences that are difficult to replicate through fully AI-driven education. Technology can be a powerful teacher. But growing up is still a human process.

https://t.co/Z790NHkDs2

I'm building something new: https://t.co/7Esv1kIPvH, my next venture in AI news. So when @_ahacreator asked me to test their 3.0 influencer marketing platform, I gave it a real job. I pasted one link, https://t.co/XkU8fN1qYo, and watched its AI build the entire influencer campaign for my own brand. What happened next genuinely surprised me.

The AI read https://t.co/XkU8fN1qYo and understood it instantly. It knew an AI news brand belongs on X, LinkedIn, and YouTube, and it deliberately skipped TikTok and Instagram. It generated creator archetypes that fit the mission: fact checkers, tech news reporters, media literacy educators. Then it matched real creators by name, each with a written reason and a fit score, backed by verified audience data. And it didn't just match. It screened for risk too: content tone mismatch, brand safety issues, audience mismatch, even past content problems. For a news brand, that filter matters as much as the match. 100,000+ active creators. 5M+ in the pool. Minutes, not weeks.

Here's the bigger picture. The whole operational layer of influencer marketing (screening, outreach, negotiation, tracking) just became a 24/7 AI agent. One person can now launch what used to take a team. That's how I plan to grow https://t.co/XkU8fN1qYo, and AhaCreator 3.0 is exactly the kind of tool making it possible. Keep an eye on https://t.co/XkU8fN1qYo. The campaign you just watched it build? You'll see it run. Try it yourself: https://t.co/3xJwlizD88 @_ahacreator

+1 more

Falcon 9 launches 24 @Starlink satellites from California https://t.co/YbjRLQBEZo

+1 more

The UK arrests people over retweets. Crazy. https://t.co/AKmfHsr6Gr

// HarnessX: Harnesses You Compile, Not Hand-Build // (bookmark it) Most agent harnesses are hand-crafted and frozen. Each new model or task means rewriting the prompts, tools, memory, and control flow from scratch, and the rich traces from every run get thrown away. HarnessX treats the harness as something you assemble from typed primitives through a substitution algebra, then evolves with AEGIS, a trace-driven multi-agent engine that feeds execution history back into the design. Why does it matter? If the scaffolding can compose and improve itself from its own traces, the per-model rewrite tax that quietly dominates agent engineering starts to disappear. This is the cleanest version yet of treating the harness as a first-class, programmable artifact. Paper: https://t.co/zvz5YTLRZr Learn to build effective AI agents in our academy: https://t.co/LRnpZN7L4c

Three things @posthog does really well on their homepage: 1. The site is an OS: windows, desktop icons, a taskbar. Commitment to the bit 2. The cookie banner is titled 'Legally-required cookie banner' 3. The copy sells to your AI agents, not just you Grabbed with one line. https://t.co/4gVLPPdQv5

WOAH! Hermes Desktop with the HTML artifact preview just blew me away 😍😍😍😍 thanks @imbabybrooklyn, really dope update <3 https://t.co/UCH8dQMJPx

Few seem to have noticed we have full live artifacts in the desktol GUI app for Hermes Agent! Check it out: https://t.co/L555n24W3T