Sci-fi irl

Beautiful view of Starlink V3 deployment

The OpenAI-Hugging Face hack was enabled by weak sandboxing. It is great that Nvidia is releasing open source tools for sandboxing AI agents. OpenWorker, our open-source agent harness supporting cybersecurity workflows, is proud to support this. A sandbox gives an agent limited permissions. OpenWorker is building on Nvidia OpenShell and will support running each agent's commands inside a sandbox. Only the files relevant to the task go in. Secret API keys, your web browser login credentials, the ability to access arbitrary websites, are inaccessible to the agent by default. These restrictions are implemented in deterministic code rather than by prompting an LLM, which can make mistakes or be susceptible to prompt injections. Further, all actions are logged for monitoring and audit. I'm grateful for @JensenHuang's leadership making AI agents more secure. OpenWorker (which @rohitcprasad and I are working on) will continue to improve security for agents.

Here's my conversation with Khabib Nurmagomedov (@TeamKhabib), one of the greatest fighters of all time. It was truly an honor to get to train with him, to have this deep conversation, and to get to know him over the past few months. We did this conversation entirely in Russian. It's translated and dubbed into English. Both language audio tracks and subtitles are available on YouTube. Here on X, I'm posting the English & Russian language versions as seperate videos. The translation & dubbing was a huge amount of work (which I deeply care about) with humans & AI collaborating. Thanks to the amazing team at @ElevenLabs for their help. Good luck to Islam Makhachev, Khabib & team this weekend at UFC 330! And as I've said in the video, I'm truly grateful for all of you, for your support all these years. I love you all ❤️ Timestamps: 0:00 - Introduction 1:35 - Why Dagestan produces so many great fighters 11:41 - History of Dagestan warrior culture 21:02 - Hard training in the mountains of Dagestan 24:53 - The trap of fame, money, and power 35:34 - The key to Khabib's dominance 46:47 - Discipline and hard work 59:15 - Street fights in Dagestan 1:12:56 - Khabib's top pressure 1:18:40 - Khabib vs Conor McGregor 1:31:25 - Khabib's toughest fight 1:39:12 - Greatest MMA fighters of all time 1:56:57 - Coaching 2:02:30 - Daniel Cormier 2:06:48 - Islam Makhachev 2:16:46 - Dana White 2:20:16 - Joe Rogan 2:22:31 - Ali Abdelaziz 2:25:08 - Football 2:32:23 - Messi vs Ronaldo 2:50:23 - Dubai 2:54:14 - God 3:01:57 - Advice for young people

This is Ultrafast. Our premium speed tier, Ultrafast offers up to 8x faster token generation (300 tokens per second) in Codex and up to 6x in the API.

Astra for Law — tools and skills to radically accelerate law firms. Includes data privacy as a core feature, and 26 partner-built plugins and 47 community plugins for legal work in ChatGPT.

Hoi Nederland! 🇳🇱 Come and join the team: elevenlabs.io/careers

The questions posed by AI are bigger than the AI community. We urgently need the world – religions, civil society, academics, governments – to participate in creating a positive outcome. I'm glad the Catholic Church is engaging, and honored to speak at the presentation.

We would love to see more collaboration and research in the field of human-AI interactivity. Check it out!

Introducing Grok Voice Transcribe 2.0. It’s the world’s most accurate speech transcription model.

Introducing Automations in Perplexity Computer. Automations are for ongoing work. They can take action in response to event-based triggers or on a schedule. Automations work with your memory, skills, and connected apps like Slack, Gmail, Outlook, and Linear.

Can our TPUs survive and operate in space? Well, we're going to find out. Project Suncatcher is hitching a ride aboard @SpaceX's Transporter-18 mission, testing a prototype satellite built in partnership with @planet One small step for TPUs....

Made this in Cowork using just this prompt and these two images: "Create a 90s animated video with a synthesized voice of an alien exploring Borges library of babel and attempting to use the library to understand human culture yet dealing with the impossibility inherent to the vast search space of language. The video should have a clinical and spooky vibe that gets at the vastness of the library and also the fundamental challenges in lifeforms understanding one another. As inspiration, also take a look at this old GUI from a film as well as this piranesi etching."

The most important thing about Sonnet 5.5 is that it's now the model that powers the free tier on claude.ai - so all of this stuff can be done by free users ChatGPT's free tier is still GPT-5.6 Luna, which is a lot less capable

I've spend 10+ hrs perfecting how to get the best out of Opus 5.5 for video generation workflow: - Use it with Claude Code, not the app. - Use OpenRouter API to access all the other models (image gen, video gen, audio gen) with one key - Use Gemini 3.8 TTS and make it create a skill from the API to put emotion into the voice. - Use Manim, Hyperframes, Motion Canvas for motion graphics for HTML web grounded video - Use GPT 2.5 Image Sunburst for keyframes and Veo 3.1 / Seedance 2.5 for generation. Seedance does better with motion shots. Ensure you have reference images for consistency and generate an animatic before full video. - Add a script writing planning skill. Use OpenTimelineIO to assemble. - Explicitly tell it to avoid Claudisms like short punchy sentences and a lot of numbers. “Narrate like a university professor.” - Allow it to use yt-dlp with search to inject clips from real videos (creative commons) in composition - Use ElevenLabs for music gen in the background - Add a caption skill which uses any decent ASR model to get word-level timed subtitles on the final video - Add a critic skill which takes screenshots and uses transcription to validation the audio and video quality In the prompt, specify a) what you want b) aspect ratio c) length d) style preferences. ffmpeg handles most of the actual video manipulation tasks. Here's an example of a 4 minute Neocloud business explainer:

annual ranking of best ai podcasts just dropped

I’ve been testing Dots for a bit. In short, it has a potential to be the best agent product in the world. Why? The quality of the AI is best-in-class. But the UX needs a lot of work before I can make it my daily driver. If OpenAI nails that, they’ve got a huge winner.

Interesting approach. Models can't output freeform text but can choose from a set of predefined values. Could be useful for data labeling and tasks with a fixed list of possible actions. Unclear how reasoning would work though, but it's super cheap (output tokens are free!)

As one example, Fo can take a request like booking a dentist for Thursday and work on it all the way through to a confirmed appointment. A free Fo Mini tier comes with a daily task limit, while an invite unlocks OG Fo with unlimited tasks and call time. Test it out 👀 wajo.ai/

The loudest voices stoking fears about AI dangers have made tremendous headway in the past two weeks. AI technology has not taken some unexpected, dangerous turn, but the hype around it — propelled by what appears to be a well orchestrated PR campaign — has drummed up considerable fear. I worry that it represents a setback for our field. I have written frequently that fears of AI are overhyped. AI’s capabilities can be uncannily human-like and unpredictable, and it’s rational to worry when people who are directly involved express concerns. But I see the problems as a sign of the engineering work that ahead, rather than insurmountable barriers or the sky falling. AI technology continues to advance — which is a good thing! — but technical advances, poorly understood by the public, give those who seek to generate hype repeated opportunities to do so. First, I don’t see any step up in the risk of human extinction from AI compared to a few months ago. The theories about this remain the same fantastical, science fiction scenarios as a few months ago. The biggest change in AI risk is its cybersecurity capabilities — a topic which we should take seriously — but this, too, will not lead to the end of the world. The most notable recent event leading to increased fear was when an OpenAI team deployed an agent swarm that hacked into Hugging Face. Much of the popular press contained significant hype. For example, some publications reported that a swarm of 1,200 agents carried out the attack. While this was technically accurate, as I write this, I have about 1,300 processes running on my laptop. Yes, the ability to get large swarms of agents to work in parallel on a task is a significant technical advance, And, in computing, many processes run at the same time. So this shouldn’t be seen as some magical capability. Additionally, OpenAI’s buggy sandboxing and monitoring processes were key to enabling this incident. Fixing these bugs and putting in place improved monitoring would be appropriate fixes, not pausing AI. There are many well known ways to attack software systems. The main advantage of AI agents is that they are relentless. They will tirelessly try many tactics — and have the patience to chain vulnerabilities together — that previously would have taken an infeasible amount of human effort. But in the long term, I believe the advantage will lie with defenders (because they have more information with which to identify bugs, which they can fix), but the cyber-threat landscape has changed significantly. There are still bottlenecks to identifying and exploiting a vulnerability. AI agents still have to try a lot of things to see what works, and taking these actions takes time and might be detected by defenders. This is why, even though it is now easy to obtain versions of leading open weight models that have had their guardrails removed or weakened, so they will not refuse to try to execute cyber attacks, the world has not ended. I am also concerned about the anthropomorphization of AI in a lot of reporting, where LLMs and agents are unnecessarily treated as if they were people. If I wield a hammer, miss a nail, and accidentally dent the wall, it’s not the fault of the hammer. The problem lies in how I used the hammer. Similarly, if I prompt an agent and it hacks into someone else’s system, the responsibility lies with me, not the agent. Of course, we want to build systems that are as safe and predictable as possible. (For example, an unsafe hammer would be one whose head randomly flies off under normal use.) Today’s agentic systems are not predictable, but I see no reason why, by applying sound engineering practices, we won’t be able to make them extremely safe to use. One new element in the forecasts of AI-enabled doom is AI companies disclaiming responsibility for their own products. “I didn’t do it; my out-of-control agent did!” There’s a balance to be struck between the responsibility of the tool maker and the tool user, but when something goes wrong, let’s hold the people building and/or using the hammer responsible, rather than the hammer. (By the way, if you’re worried about AI bioweapon risk, David Bellamy has a great post on why this, too, is overhyped. Briefly, the bottleneck in building a bioweapon is not intelligence, but lab work and manufacturing.) Pausing AI progress will create much more harm than benefit. First, our adversaries will certainly not slow down. Second, engineering requires discovering problems empirically so we can fix them. If we pause AI by a decade, we will also delay finding and implementing safety engineering fixes by about the same duration. Of course, the incentive to stoke fears — for regulatory capture, to garner attention, or to make one’s technology seem more powerful — remains the same as before. Disclaiming responsibility is a new one. Taking a hard technical look at the actual risks however, I see little factual basis for the degree of fear that’s been stoked up. We still have hard research and engineering work ahead to improve AI safety, but the beneficial applications continue to vastly outweigh the risks, and we should keep building. [Original text (with links): deeplearning.ai/the-batch/is… ]