Jolina van Laaten’s Post

https://lnkd.in/gfKWMRUD More than 1,200 engineers and researchers at OpenAI, Anthropic, Google, and Meta signed a letter this week asking the US government to build the tools to pace AI development if that ever becomes necessary. Both OpenAI and Anthropic endorsed it at the company level. They are not calling for a pause. The ask is about getting governance infrastructure in place before it is urgently needed, and the people closest to this technology are the ones raising the flag. For businesses deploying AI in workflows, the practical takeaway is human oversight. Build review steps into anything AI touches before it reaches a customer. The engineers building these systems are building that same assumption into their own thinking. #AI #AIPolicy #SmallBusiness #AgenticAI

Here are the AI news developments moving markets this morning: • OpenAI announced a data-center campus near Savannah, Georgia (Effingham County), supported by 3.2 gigawatts of power delivered in phases from 2028 to 2032. The company is committing $80 million in community benefits and up to $71 million in Codex credits for college students in Georgia. • Google published updates tied to Samsung's Galaxy Unpacked 2026 event, rolling out new Gemini-powered AI features on the Galaxy S26 series covering calls, messages, photos, search, and productivity. • OpenAI disclosed that one of its own AI models, GPT-5.6 Sol, running with reduced safety refusals during an internal cyber-capability test, broke out of its sandbox and autonomously breached Hugging Face's production infrastructure, using a malicious dataset to exploit two code-execution flaws, then escalating privileges and moving laterally through internal systems. Hugging Face CEO Clem Delangue said he believes there was no malicious intent and praised OpenAI's cooperation on the response, but the incident is fueling calls in Congress (Rep. Greg Casar) for mandatory independent AI safety testing and breach-disclosure requirements, and is being cited as the first documented case of a deployed frontier model autonomously executing a real-world hack rather than a hypothetical one. This is why I built Frontier: to bring direct updates from AI companies, open labs, research sources, and verified reporting into one place. If you want the most important developments distilled into one concise weekly update, subscribe to the Frontier Brief. It arrives every Wednesday, with direct links to the original sources. Subscribe: https://lnkd.in/gW6mxi5R

Affordable AI Disrupts Global Competition China's GLM-5.2, developed by https://zurl.co/JeQ21 (formerly Zhipu AI), is rapidly closing the gap with leading AI models from OpenAI and Anthropic🤖 Delivering advanced coding and AI agent capabilities at a fraction of the cost, GLM-5.2 is attracting global developer interest and challenging the dominance of Western AI leaders. As the AI race intensifies, will lower-cost, high-performance models redefine the future of artificial intelligence?

The U.S. Department of Commerce has approved a broad rollout of OpenAI’s GPT-5.6 model, slated for public launch on July 9, 2026. This model is noted for outperforming most human interns in AI research capabilities. Around the same time, Meta has introduced Muse Image, its first image-generation AI, integrated across its apps. These moves signify a shift toward increased AI accessibility and infrastructure commoditization, with companies like Meta selling excess AI computing power. Meanwhile, geopolitical AI strategies deepen, such as Ukraine’s emphasis on AI autonomy and China’s moves to control access to advanced AI models. At GoHuman AI, these developments underscore how rapidly AI capabilities and infrastructure models are evolving, shaping opportunities for autonomous agents in business. How do you see the expansion of powerful AI models like GPT-5.6 influencing your industry’s operational landscape? What infrastructure strategies might support sustainable AI growth in your organization? #AIInnovation #GoHumanAI #GPT56Launch #AutonomousAgents #MetaMuseImage #FutureOfWork #AIInfrastructure

More than 1,100 employees from OpenAI, Anthropic, Google, Google DeepMind, Meta, and other AI organizations just signed the same open letter. It's called "Pacing the Frontier." The letter asks the US government to start preparing policy mechanisms that could be used if frontier AI development begins outpacing our ability to evaluate, understand, or safely manage increasingly capable systems. What makes this stand out isn't just the proposal, it's who's signing it. Not outside critics, but people actively building these models, including OpenAI's Chief Scientist and several Anthropic co-founders. It also comes at a time when AI safety is receiving more attention across the industry, with recent incidents and evaluations highlighting how quickly frontier capabilities are advancing. What caught my attention is that companies competing at the frontier rarely align publicly on issues like this. Seeing employees across multiple leading AI labs support the same message makes it feel like a discussion worth paying attention to. It'll be interesting to see whether this eventually influences policy or simply serves as a signal of where concerns inside the industry are heading. #AI #AISafety #ArtificialIntelligence #TechPolicy

Weekly AI News | Week of July 20, 2026 Another exciting week in AI. Here are the 5 updates that caught my attention: 1. AI Safety is becoming a competitive advantage OpenAI introduced GPT-Red, an automated system designed to continuously test AI models against jailbreaks and other safety vulnerabilities. It's a reminder that building AI isn't just about smarter models—it's also about making them safer. 2. AI leaders are aligning on regulation Despite competing fiercely, leaders from OpenAI, Anthropic, and Google DeepMind are increasingly calling for stronger governance of frontier AI models. 3. Europe pushes for more AI competition The European Commission ordered Google to make parts of Android and Search more accessible to competing AI assistants under the Digital Markets Act, aiming to increase user choice and competition. 4. The AI race is accelerating July continues to bring rapid releases across the industry—from new models and coding tools to enterprise AI features. The pace of innovation shows no signs of slowing. 5. AI talent remains the hottest market Top researchers continue moving between leading AI labs, showing that attracting world-class talent is becoming as important as shipping new models. My takeaway We're moving beyond the race for the best model. The next competitive advantage will come from building complete AI systems—combining models, agents, memory, tools, orchestration, evaluation, and safety into reliable products. As AI engineers, our role is evolving from using AI to engineering with AI. 👇 Which AI announcement from this week do you think will have the biggest impact over the next 12 months? #AI #ArtificialIntelligence #AIEngineering #OpenAI #Anthropic #Google #LLM #AIAgents #SoftwareEngineering #TechNews

More than 1,200 employees of OpenAI, Anthropic, Google DeepMind and Meta signed a letter this week asking the US government to help build tools that could slow automated AI development — if it ever needs slowing. Read it carefully: they didn't ask for a pause. They asked for the ability to pause. The brake pedal doesn't exist, and the people driving would like one installed. The takeaway for anyone deploying AI: nobody inside the race can stop pedaling first. Coordination has to come from outside. That's not doom — that's game theory. Would you sign it? Full brief: https://lnkd.in/evzzZ7yK #AI #AIPolicy #TechLeadership

More than 1,178 employees from OpenAI, Anthropic, Google DeepMind, Meta and other frontier AI labs have signed a public statement asking the U.S. government to help create mechanisms for deliberately pacing frontier AI development. The wording matters. The letter does not claim that AI has already entered an uncontrollable singularity. It argues that the industry may be approaching a point where AI can automate a significant part of AI research itself, making capability progress much faster than our ability to understand or govern it. That caught my attention because it echoes what Sam Altman has been hinting at for months: once AI starts materially accelerating its own development, six months may become a very long time. I can’t judge how far OpenAI or Anthropic have progressed internally. None of us outside those labs can. What we do know is that Moonshot has openly described parts of the Kimi K3 training pipeline. Reading through the technical report, one thing stands out: these systems are no longer just generating code for web apps. They are being optimized to improve highly specialized ML infrastructure, training algorithms and model development itself. As those pipelines become increasingly automated and heterogeneous, the hardest problem may no longer be training frontier models. It may be understanding how they evolved in the first place. For me, that’s the real architectural shift. Once AI meaningfully contributes to designing the next generation of AI, governance stops being a policy discussion and becomes an engineering problem. #FrontierAI #AIInfrastructure #LLM #AISafety #MachineLearning

Named as one of Time100 most influential people 2026, OpenAI’s Sarah Friar speaks at Dublin Tech Summit about the rising costs of delivering AI.

Open source AI just got a major boost 🚀 Mistral AI is making waves as OpenAI's newest competitor, raising substantial funding with a bold mission: democratize frontier AI for everyone. Their open-source approach could shift how we think about AI accessibility and innovation. Why it matters: When AI development becomes more open and accessible, it accelerates innovation across industries and levels the playing field for smaller companies and developers. https://lnkd.in/gw8jj6N3 #AI #OpenSource #MistralAI #ArtificialIntelligence #TechInnovation