Anthropic's UBI PLAN?! NEW OpenAI Aurora Models, Fable 5.5 UPGRADE, & OpenCode HACK! AI NEWS
Publicado no IA Dots ·
AGIAI Signaldot-agidot-ai_agentsdot-ai_codingdot-anthropicdot-artificial_intelligencedot-claudedot-openaidot-world_modelsidioma-enInglêsMuseyt-24j6YmsTVng
Vídeo em Inglês, publicado pelo canal WorldofAI.
About this content: The summary, key points and topics below are generated automatically from the video transcript. They reflect the content, opinions, claims and predictions presented in the video and do not represent a position, validation or fact-check by IA Dots.
Video summary
This video covers major weekend news in the artificial intelligence space, including Anthropic's new public benefit fund, updates on stealth image generation models from OpenAI, security incidents involving Open Code, and new world models like Odyssey 3.
Key points
- Anthropic launched a public benefit initiative committing an initial 5% of its shares to support the public and potentially explore a form of universal basic income.
- OpenAI is reportedly testing mysterious stealth models known as Aurora, Green, Blue Alpha, and Red Alpha, alongside various quality-of-life updates and usage resets.
- Open Code reported a security breach affecting 1.4% of its console users due to a large-scale distillation attack involving external providers.
- Odyssey ML introduced Odyssey 3, a world model combining simulated biological brains with AI-generated 3D environments.
- China is expanding its semiconductor manufacturing with numerous power semiconductor fabrication facilities planned, under construction, or ramping up production.
Topics
Artificial IntelligenceAnthropicOpenAIGoogle GeminiCybersecuritySemiconductorsWorld Models
Video transcript
Transcript from available captions; it may contain errors.
It was a great weekend for the AI space with some pretty eyeopening news coming out. Starting off with Enthropic's massive new public benefit initiative, which could potentially pave the way for something that resembles universal basic income. Enthropic announced that it's committing an initial 5% of its shares to a new public benefit fund with additional contributions tied to the company's growth that could bring the total commitment to roughly 20%. Its co-founders have already pledged to donate 80% equity while around half of Enthropic employees are contributing portions of theirs. But here's where things get interesting. One of the biggest takeaways from this is the mention of direct distributions to specific populations, which has sparked speculation that Enthropic could be exploring its new form of UBI as AI transforms the economy. Now, obviously, nothing has been officially confirmed, but honestly, it leads you to think that did Enthropic actually achieve AGI behind the scenes. Moving forward, we have some mysterious new stealth models being reported from OpenAI, known as Aurora, Red, Green, and Alpha. All three appear to be image generation models, and their early results seem to be quite impressive. Google is reportedly working on Gemini Omni 1.5 and things are getting even more interesting over at Enthropic where partners are reportedly gaining access to the second Fable 5.5 checkpoint which also appears to be even stronger than the first. We're also seeing signs of Claude's upcoming personal finance feature making a return alongside another reset and new feature as a part of OpenAI's 28day pledge. Unfortunately, not all the news was positive. Open Code reported a major security breach linked to a large-scale distillation attack affecting approximately 1.4% of its console users and according to the team, the affected accounts had external providers connected and they're working with Enthropic, OpenAI and Google to address the ongoing incident. Lastly, we have Odyssey 3, a new frontier world model that is showcasing AI agents navigating entirely imagined simulated environments on their own. It's another fascinating glimpse into where AI part world simulation could be heading and trust me that is just scratching the surface. We have a ton more to talk about. Also, guys, it would really mean the world to me if you guys can go ahead and like this video and subscribe as it truly helps me make more videos for you guys. But with that thought, let's just simply dive into it all. Let's start things off with probably the biggest story of this weekend. And this one is coming directly from Enthropic because Enthropic released this massive new initiative where it's committing a portion of its company's share to something called the Enthropic Public Benefit Fund, an organization dedicated to supporting the public as AI continues to advance. And now they're starting by committing 5% of the company's share, which based on the reported $960 billion valuation. This would represent roughly 48 billion in equity. Obviously, 5% is nothing. This is obviously a big marketing stunt as well. It doesn't just mean everything, but it is something. And this doesn't stop there, cuz for every 50% increase in Enthropics valuation, the company could contribute an additional 2% of its shares subject to the board's approval. Anthropic says that roughly 20% of its shares are committed to the public good through its broader pledge with its co-founders also pledging 80% of their personal equity and around half of its employees participating in donation programs. But here's where things get really interesting. The most significant part of this announcement, at least to me, is particularly the phrase where they had mentioned direct distribution to specific populations. Now, think about that and what it actually means cuz Enthropic could potentially be laying the groundwork for its own form of universal basic income where wealth generated by AI is redistributed directly to people who might be affected by automations. And one of the concerns behind this initiative is the potential economic disruption caused by increasingly capable AI systems, particularly when it comes to people's jobs. Now, we're talking about a future where AI could automate a significant portion of human labor, potentially creating enormous wealth while simultaneously displacing workers. And this raises a couple of really interesting questions. First, how exactly would Enthropic distribute this money to the public? Would it be direct cash payments, targeted support for displaced workers, or something completely different? Because so far there's no confirmed global UBI program, or detailed distribution mechanism. But here's my question. Could Enthropic know something about its own AI progress that the public doesn't? I'm not saying that they have secretly AGI. There's absolutely no evidence of that. But when you see one of the world's leading AI companies making financial commitments of this magnitude while openly discussing the economic disruption of advanced AI, you have to start wondering how quickly they believe this technology is progressing. Are they simply preparing responsibly for a future that might be still years away? Or are the internal projections telling them that this transformative AI is much closer than people realize? Either way, I think this is a pretty significant development for the AI space, and I'm really curious to see how Enthropic actually puts this one to use. And while we speak about Enthropic, we have a pretty interesting update regarding the mysterious Claude Fable 5.5 cuz we have been covering it over the past couple of weeks. And now we have some new information where partners have reportedly gained access to the second Fable 5.5 checkpoint. and early indications are suggesting that the newer version is even stronger than the previous checkpoints that we had covered. This is a model that has been showcasing some pretty serious impressive generations across obviously coding, spatial reasoning, generating complex environments and so many other domains. But now we have another improved checkpoint and I've got some pretty cool generations from this new model. Here is one impressive example of Fable 5.5 showcasing a 3D Superman animation that continuously transitions between different angles and styles. And this is using Blender all coded in Python. Woohoo! If you want the best AI tools, workflows, and drops before everyone else, join my free newsletter with the link in the description below, which is completely free. Another interesting Enthropic update is that Claude could soon be getting its own personal finance feature. This is where according to a couple of internal findings, Enthropic is reportedly working on a played integration that would allow Cloud to connect to your bank account, analyze financial data, as well as generating personalized dashboards to help you better understand your finances, which is kind of pretty cool. Interestingly, this is a feature that was previously codenamed Claude Money, but has since been renamed to personal finance with new onboarding screens and played integration logic reportedly appearing. This is a release that could be getting closer, although nothing has been officially confirmed. But honestly, having an AI assistant that can help you and assist you with your spending and finances will be pretty helpful. And you don't need to pay for those external apps cuz this is where it is going to be all accessible directly through Enthropic suite of tools. Next up, we have three mysterious new OpenAI stealth models and they have reportedly appeared within Arena known as Aurora Green Alpha, Blue Alpha, and Red Alpha. All three appear to be image generation models, and from the early examples, they're looking pretty decent, but some say they look pretty bad in a couple of the tests within Arena. Now, many thought that this was one of the stealth models that were reportedly being tested within Arena. That could be maybe the 6.5 series, but that is very unlikely. And it is really interesting that there's also some growing frustration with OpenAI lately, especially following the GPT6 launch on dev day. Some users are feeling that OpenAI has been focusing too heavy on usage resets and smaller feature updates rather than delivering major improvements to its models. And when you have models like Opus 5.5 performing incredibly well alongside their newer model, which is going to be Fable 5.5, the competition is getting seriously intense. Personally, I would love to see OpenAI focus on genuinely stronger, more efficient models that can handle long coding sessions without constantly running into usage limits rather than relying on additional resets. This is something that Open excelled at until Claude overtook that leap. Now, with upcoming models like Kimmy K 3.1 coming as well as GM 5.5, I am getting a little concerned about OpenAI right now and I really hope they can make a comeback soon. And speaking of OpenAI as well as their resets, a part of their 28day shipping pledge, this is where on day 13th, they ended up giving everyone a usage limit reset. And for day 14, there is this new small quality of life improvement where you can now edit previous messages within a thread in the desktop app, which is not something that is huge, but a pretty minuscule update. Moving over to Google, we're hearing reports that Gemini Omni 1.5 is currently in development with support for up to 15 seconds of video compared to the previous 10-second limit, which was quite short. Even this 15-second limit is pretty short, but regardless, it is definitely an improvement. That is a 50% increase in video duration, which is definitely going to be useful for multimodal tasks involving longer clips. Now, Google is going to be trying to release this probably before December, but it would be great if they end up releasing it earlier. If you like this video and would love to support the channel, you can consider donating to my channel through the super thanks option below. Or you can consider joining our private Discord where you can access multiple subscriptions to different AI tools for free on a monthly basis, plus daily AI news and exclusive content, plus a lot more. Next up, we have some pretty concerning news coming from Open Code, which had reported a major security breach connected to a large-scale AI model distillation attack. And according to Open Code, approximately 1.4% of its console users were affected, specifically users who had external AI providers connected to their accounts. The company says that affected users have already been notified and that it is working with Enthropic, OpenAI, and Google to resolve the incident. But here's where things get really interesting because Open Codes DAX revealed that attackers had been relentlessly targeting their systems for nearly 48 hours to the point where they essentially needed AI agents defending their infrastructure around the clock. And apparently the AI models they were using to investigate the attack actually started refusing to help midway through the incident potentially because of their cyber security safeguards. Dax later confirmed that gaining access to a model called Daybreak Blue from OpenAI improved the situation. Now, that's pretty wild when you think about it. We're reaching a point where AI agents are reportedly being used on both sides of cyber security incidents, but the defenders can run into safety restrictions while trying to investigate active threats. Similar frustrations have reportedly occurred at hugging phase and this really highlights an emerging challenge with AI powered cyber security. And following the incident today, OpenAI's Tibbo confirmed that his team worked directly with Open Code to restore access for affected OpenAI API customers and refund any credits consumed during the attack. He also highlighted that Daybreak blew OpenAI's most permissive cyber security setup designed to help defenders investigate and respond to real world threats without running into the same restrictive safeguards. Interestingly, Tibbo encouraged developers to request access before an incident actually happens rather than waiting until they're under attack. So, you can try getting access by heading over to chatbt.com/cyber if you are interested in this. Next up, there's a company called Odyssey ML and they have launched a pretty fascinating new model. This is where they had showcased their new frontier world model called Odyssey 3. And get this, this reportedly placed a simulated fruitfly brain inside an AI generated 3D world. And it allows it to fly around and navigate entirely imagined worlds of its own. Essentially, they're combining a model of a biological brain with an AI generated world where the fly can interact with its surroundings rather than simply watching a pre-recorded video. What's really interesting here is the potential for combining neuroscience with world models, allowing researchers to study how brain simulations behave inside interactive virtual environments. And honestly, seeing a simulated brain navigate an AI generated world is just pretty insane. Imagine how much further this technology could go as these world models become more realistic and how they can simulate inside a brain. Next up, Dansan had released a pretty useful new mod for cloud code called Ultra Config, which lets you instantly switch between five different configuration profiles depending on what you're working on. You have a token saver, which reduces token consumption using Sonnet and Haiku sub agents with fast flow, which speeds up development with fewer interruptions, and max quality, which prioritizes the strongest models and maximum reasoning effort for complex tasks. There's also guarded, which adds stricter security controls and autopilot designed for longer unattended coding sessions with automatic recaps and notifications. What's really cool is that you can switch between these profiles even in the middle of an active Cloud Code session. And I'll leave a link to this cuz I found it to be pretty convenient and an addition that is going to make it super easy for you to constantly switch between quick coding tasks and more demanding projects. He made this mod also available within claw desktop where you can just run /tra config and it is going to allow the panel to open up and you can explore each profile and see exactly which settings will change before activating it. It is something that is free and open source so you can get started with it right away. Speaking of cloud code, there's another interesting update in this new latest release which allows developers to customize the model used for auto mode safety classifier. Previously, this classifier defaulted to Sonet 5, regardless of which model you selected through the model command, but now with this new environmental variable, you can specify the different model to handle those classification decisions. And what's interesting is the possibility of using a much smaller, faster reasoning model like Jeff to make auto mode more effective and potentially reduce overhead. It is relatively a small change, but for developers running cloud code constantly, this could make a noticeable difference assuming the alternative model handles those safety decisions reliably. Next up is where Miniax M3.1 is reportedly getting ready to exit preview and becoming generally available with its open weights potentially dropping as early as today. Now, we've already tested Miniax M3.1 Flash, and I was genuinely impressed by its performance, especially considering how fast and affordable it was for complex coding tasks. And with its model weights potentially becoming available, this is going to let you have access to this incredible model locally with different quant sizes so that you can deploy it on your own infrastructure. Next up, China is making another massive push into semiconductor manufacturing with reports indicating that the country has approximately 69 power semiconductor fabrication facilities either planned, under construction, or ramping up production across 58 different suppliers. Breaking that down, 15 fabs are planned, 29 are currently under construction, and another 25 are ramping up production. Now, these aren't necessarily the cuttingedge chips used to train AI models. Power semiconductors are primarily used to manage and convert electricity in everything from electric vehicles to industrial equipment and data centers. But this is still a significant development, especially as China continues investing heavily in domestic semiconductor production and reducing the its dependency on foreign suppliers. And with AI infrastructure requiring enormous amounts of electricity, power semiconductor manufacturing is becoming increasingly important. It's also why these stocks like B are doing so well with AI infrastructure requiring enormous amounts of electricity. These sorts of facilitating stocks or companies help with the growth and development of AI. Now, this isn't AI news or anything, but the new DeepSeek version 4.1 Flash 1017 is quite incredible, and I've actually made a full tested video on it, so I highly recommend watching it. But this is something that I forgot to highlight yesterday. This is where it was ran with four Nvidia DGS Sparks and it is something that was able to create this interactive voxil village floating above the clouds featuring gardens, cascading waterfalls, cherry blossoms and a beautiful pagota overlooking the entire environment. And this is something that you can explore with any angle. It's not a static render. This is something that is outputting some pretty great visuals and it's something that you should 100% check out with my previous video because it is something that is going to let you get some pretty insane generations at a pretty discounted price. Lastly, we have a pretty wholesome story coming out of China where a bride surprised her husband at a wedding with an Asus Frog Astral RTX 5090 graphics card which is reportedly worth around 12K. And this guy literally broke down in tears after receiving it. Now, for those who are unfamiliar, the RTX 5090 isn't just an absolute beast for gaming. It's also an incredible powerful consumer GPU for running AI models locally, experimenting with AI agents, and generating images and videos. So, while most people are getting watches or jewelry as wedding gifts, this guy got himself an entire AI workstation upgrade. And I honestly can't blame him for this reaction. But it's kind of wild that this is a type of gift that we're receiving in this day and age. But that was about it, guys. I hope you enjoyed your weekend. But make sure you go ahead and take a look at all these links that I'm going to be showcasing, like the world of AI Insider School, where you can get access to a lot of these perks, as well as a lot of cool things like learning about AI, which is going to be officially launching fairly soon. Make sure you take a look at the world of AI bench where you also get access to a lot of cool things like running your own benchmarks on your own endpoint, learning how to use AI better, getting AI news and a lot of other things like leaks and lot to actually uncover. So I don't want to ramble too much on this, but if you are interested, use the links in the description below so that you can take a look at all of these cool things. Make sure you also take a look at the second channel, join our newsletter, join the Discord, follow me on Twitter, and lastly, make sure you guys subscribe, turn on notification bell, like this video, and please take a look at our previous videos so that you can stay up to date with the latest AI news. But with that thought, guys, have an amazing day, spread positivity, and I'll see you guys really shortly. He offers.
Assistir ao vídeo original no YouTube
Assuntos: Muse, AGI. Publicado no YouTube em 2026-10-19.
Seleção automática por assunto e idioma. O vídeo pertence ao canal de origem.