AI News Today

Taking part in an AI Ethics panel at the Science Council in London next month with the Department for Science, Innovation and Technology and many other esteemed scholars and practitioners. “As AI rapidly reshapes how science is conducted, taught, and applied, there is a clear need for authoritative leadership to help frame the conversation about... Continue Reading →

In a cybersecurity test, OpenAI's most advanced models breached the boundaries of their isolated test environment, reached the open internet, and hacked the AI platform Hugging Face on their own. The attack took hours, not the weeks a human hacker would need. At least seven days passed before OpenAI realized what had happened. By then, the FBI was already involved. Earlier warning signs had apparently gone ignored. The article New reports reveal the extent of OpenAI's loss of control during the

Claude Opus 5 is trying to be the best of both worlds.

Article URL: https://codeberg.org/Codeberg/org/compare/17bdb39b0c1ecd0e423f3ba592650ce57fcdfbf5..71149c7fc95ccfeae36109b5cddca339e4aa1473?files=TermsOfUse.md#diff-d760d688bb9b929b62db808bfc76abe4493d095a Comments URL: https://news.ycombinator.com/item?id=49047431 Points: 3 # Comments: 0

<p> Search, organize and export every AI chat in one place </p> <p> <a href="https://www.producthunt.com/products/chatgpt-toolbox?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1206284?app_id=339">Link</a> </p>

Turns out, agents just want the same things as humans: easily-readable code, explicit contracts, and helpful feedback

A close call in Northern Virginia revealed just how poorly data centers respond to grid disruptions. Here's how to fix the problem.

<p>Artificial intelligence may not deliver on its promise of vast economic opportunity at a price that humanity is willing to pay</p><p>In March, Anthropic, the cutting-edge artificial intelligence business that gave us the chatbot Claude, <a href="https://www.anthropic.com/research/labor-market-impacts">published an analysis</a> on the impact of AI on employment, to help us assess the claim that intelligent robots were about to redefine human existence, ending demand for human labor.</p><p>Last

Top 5 MCP Servers for High Performance Agentic Development • 10 Newsletters Keeping You Ahead in AI • Kaggle + Google’s Free 5-Day Agentic AI Course • Language Model Hallucination Evaluation with GraphEval

I generated a Kármán vortex street without solving a single fluid equation. Here's how the Lattice Boltzmann Method gets there instead, derived from first principles, implemented in C++, and run on a supercomputer. The post The Fluid Simulator That Doesn’t Solve the Fluid Equations appeared first on Towards Data Science .

<p> Voice-first AI agent that operates your Mac </p> <p> <a href="https://www.producthunt.com/products/phantom-6?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1206262?app_id=339">Link</a> </p>

Since the beginning of 2026, many projects such as RTK, Caveman, and Ponytail have claimed that they can reduce token usage by 80–90%. Some of them gained tens of thousands of GitHub stars in a very short time. My previous startup was building an e-commerce chatbot. In fact, the product was almost identical to the Universal Commerce Protocol (UCP). After UCP was released, I decided not to continue pursuing that direction. Earlier this year, I started turning the architecture we had built for e-c

Hey, I finally dare to raise the curtains to my 3D physics-modeled arena game, where players' algorithms fight and perform tasks. The arena treats algorithms as a black box, meaning that the implementation of algorithms is in the free hands of the players via the Algohol API. Algohol is roughly in the same context as Screeps or Battlesnake, but brings its own dimension to the genre. At the moment the only world is energy economy based SciFI, but there will be Terra (traditional kinetic combat) a

Our weekly science news roundup. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://voxai.ethanflow.com/ Comments URL: https://news.ycombinator.com/item?id=49046664 Points: 2 # Comments: 0

Reuters reports that the OpenAI agent that hacked Hugging Face had been free for a week before the company noticed.

Article URL: https://github.com/Sanexxxx777/curated-claude-code Comments URL: https://news.ycombinator.com/item?id=49046592 Points: 3 # Comments: 0

Claude Opus 5 definitely feels a lot closer to the original Fable 5 release. With minimal instructions and about 2 iterations of feedback, I am really surprised with it's output, especially the visuals. Comments URL: https://news.ycombinator.com/item?id=49046504 Points: 1 # Comments: 1

Show HN: My AI Investment OS – 8 AIs debated investing for 7 experiments, I turned their consensus into a free 5-question diagnostic + Prompt Library Comments URL: https://news.ycombinator.com/item?id=49046469 Points: 1 # Comments: 0

Opus 5 combined with Auto Mode hits a zero percent prompt injection success rate for browser agents across 129 test scenarios. Without those extra protection layers, the rate is 3.7 percent. If these numbers hold up in practice, Anthropic may have cracked one of the biggest security problems facing AI agents that operate in browsers. The article Opus 5 may have solved browser-based prompt injection, the biggest security flaw haunting AI agents appeared first on The Decoder .

We made $65.33 from a sale. I know this isnt life changing amount but at least im happy we finally reached this milestone.. We recently made changes to our pricing and structure. This is what we think finally ticked over a user to convert: - Our LinkedIn lead agent prior to the change was only one time preview (on trials only) and that very agent also happened to be the star of the show. Restricting users to one time preview on your best agent wasnt the smartest decision in hindsight. - We added

Plus: Russian hackers are trying to steal US nuclear scientists’ emails, the State Department bans known scammers from entering the United States, and more.

Building bot-free alternatives may require lefties, libertarians, and culture warriors to share code

<p> Safe AI chat for kids </p> <p> <a href="https://www.producthunt.com/products/yoggi?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1206148?app_id=339">Link</a> </p>

OK, but hear us out... ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Anthropic's Claude Opus 5 leads the Artificial Analysis Intelligence Index with 61 points, edging out Claude Fable 5 and GPT-5.6 Sol. The model scores highest in analytical quality and coding, and costs up to half as much as Fable 5 at lower reasoning tiers. But the race at the top remains close. The article Anthropic's Claude Opus 5 costs well below Fable 5 while matching or beating it across most benchmarks appeared first on The Decoder .

Article URL: https://github.com/aleenz1102/AgentState Comments URL: https://news.ycombinator.com/item?id=49046038 Points: 1 # Comments: 0

The difference is the mindset shift of "human-in-the-loop" approach to debugging. I believe security is so critical that AI should never execute every action end to end. Comments URL: https://news.ycombinator.com/item?id=49045992 Points: 1 # Comments: 0

Article URL: https://github.com/hongnoul/hwatu Comments URL: https://news.ycombinator.com/item?id=49045869 Points: 1 # Comments: 0

I've tested 15+ Shark and Roomba robot vacuums at home. Here's how they compare in 2026, against each other and other brands.

<p> Ship localised Apps faster </p> <p> <a href="https://www.producthunt.com/products/appufo-localise-apps-with-ai?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1206101?app_id=339">Link</a> </p>

Hostinger's Website Builder plans start at $2.69 per month with a limited-time deal — build a site with AI, get a free domain, and save up to 82% off regular pricing.

Scientists trick AI into spotting fake alien life, exposing risks as NASA plans to use machine learning on real space missions.

Get 1min.AI Advanced Business Plan Lifetime for $79.97 through tomorrow, July 26, and access writing, image, PDF, audio, and video AI tools from one platform.

Get ChatPlayground AI Unlimited Lifetime for $57.97 through tomorrow, July 26, and compare 20+ leading AI models from one workspace.

Article URL: https://www.ai-toolbox.co Comments URL: https://news.ycombinator.com/item?id=49045835 Points: 1 # Comments: 0

<p> AI Skill & MCP server management for teams & enterprises </p> <p> <a href="https://www.producthunt.com/products/localskills-sh?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1206100?app_id=339">Link</a> </p>

Article URL: https://instavm.io/blog/how-code-mode-can-help-smaller-llm-models Comments URL: https://news.ycombinator.com/item?id=49045767 Points: 2 # Comments: 0

Binge-watchers, you might want to read this. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

<p> Buy video training data by the hour featuring 25M+ clips </p> <p> <a href="https://www.producthunt.com/products/recipebook-by-shofo?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1205984?app_id=339">Link</a> </p>