AI News Today

So the idea is really simple. If CC need to change a file and run some test, CC needs to: # Turn 1 — apply patch (change package file) # Turn 2 — apply patch (fix the bug) # Turn 3 — apply patch (edit the testing script) # Turn 4 — build # Turn 5 — run tests and lint # Turn 6 — *and if is playwright CC needs another round to read media The concept is that we can use macro command to put the entire process into a RAG in one single turn. For example like this: step1: inspect env step2: apply patch

Enjoy our weekend guide to the best indie games worth checking out.

The government is piloting a program that uses AI for insurance-coverage decisions.

Article URL: https://data.stackexchange.com/stackoverflow/query/1953768#graph Comments URL: https://news.ycombinator.com/item?id=48956949 Points: 50 # Comments: 30

<pre style='white-space:pre-wrap;width:81ex'>improve(ui): settings cleanup batch — universal config.changed emitter, synced chat prefs, dedupe and copy fixes (#110581) * fix(gateway): emit config.changed from the reload committer for every accepted write Agent config_set, CLI, and doctor writes reach the gateway through the file watcher's reload path, which never broadcast config.changed - only direct RPC writes did. Move the broadcast to a single onConfigCandidateCommitted hook in the reload co

Article URL: https://velvetshark.com/ai-company-logos-that-look-like-buttholes Comments URL: https://news.ycombinator.com/item?id=48956924 Points: 149 # Comments: 38

<p>Thirty days ago, I made a bet with my team: <strong>replace our entire DevOps pipeline with AI agents</strong> and see what happens.</p> <p>No manual deployments. No human monitoring. No on-call rotations. Just AI.</p> <p>Here's the honest, unfiltered results.</p> <h2> The Setup </h2> <p>Our stack:</p> <ul> <li> <strong>Backend:</strong> Python/FastAPI on Kubernetes</li> <li> <strong>Frontend:</strong> Next.js on Vercel</li> <li> <strong>Database:</strong> PostgreSQL on AWS RDS</li> <li> <str

<p>MCP is becoming the default answer to every enterprise-agent architecture question. Access internal data? MCP. Call company APIs? MCP. Two agents need to communicate? MCP. Build the internal AI platform? Put an MCP server in front of everything.</p> <p>An MCP server is not an AI platform.</p> <p>It does not decide which source is authoritative. It does not preserve business transactions. It does not establish whether an employee may authorize a payment. It does not execute a process that wait

<p>Last night, SpaceXAI dropped <strong>grok-build</strong> on GitHub. Within 24 hours, it hit <strong>17,000+ stars</strong> and trended #1 globally.</p> <p>I downloaded it, compiled it, and spent 4 hours testing it. Here's my honest take.</p> <h2> What is grok-build? </h2> <p>It's a <strong>terminal-based coding agent</strong> with a fullscreen, mouse-interactive TUI. Think Claude Code, but with a visual interface rendered entirely in your terminal.<br> </p> <div class="highlight js-code-highl

<p>Most multi-agent demos look impressive on stage. Then they hit production and fall apart.</p> <p>Here's the pattern: agents that "worked" in a Jupyter notebook start conflicting, retrying infinitely, or silently failing when other agents are involved.</p> <p><strong>The root cause isn't the LLM. It's the orchestration layer.</strong></p> <h2> What Actually Breaks </h2> <ol> <li> <strong>No structured handoffs</strong> — Agents pass messages as raw strings. Context gets lost. Intent gets misre

Article URL: https://charlesazam.com/blog/fable-5-gpt-5-6-sol-goal/ Comments URL: https://news.ycombinator.com/item?id=48956879 Points: 7 # Comments: 3

OpenAI will send a notification to parents with linked teen accounts if their children have violated its policies around violence.

<p>This continues my "Claude Code environment" series that started with <a href="https://zenn.dev/bokuwalily/articles/statusline-live-limits" rel="noopener noreferrer">showing rate limits live in the status line</a>. This time it's a one-shot model-migration job — the <strong>"throwaway job" pattern, where the job removes its own plist with <code>launchctl unload</code> once it's finished</strong>.</p> <p>Fable 5 shuts down on 2026-07-07. If your Claude Code <code>settings.json</code> has <code>

<p>Agentic applications can fail mid-run: a network call drops, a server process crashes, an LLM request times out. Without durability built in, these failures mean lost progress and repeated (wasted) LLM calls when the process has to start over.</p> <p>Pydantic AI provides a native integration with <strong>Temporal</strong>, a durable execution engine. Temporal records each step of a workflow's execution as an event history, so if a process crashes, it can resume from where it left off instead

<p>There is a specific moment, the first time you slip on a pair of AI smart glasses, when the world acquires a faint second skin. The lenses look ordinary. The frames are heavier than the acetate you are used to, but not by much. A small LED on the rim glows for a second and then settles into something almost imperceptible. You catch your reflection in a shop window and you look, more or less, like yourself. And yet the air around your face has changed. Somewhere between the bridge of your nose

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.

<pre style='white-space:pre-wrap;width:81ex'>fix(ui): cap sidebar child sessions with show-more, drop Subagent prefix in tree (#110604) * fix(ui): cap sidebar child sessions at four with show-more, drop Subagent prefix in tree * test(ui): split child-session cap cases under the max-lines budget</pre>

<h2> What Changed </h2> <p>Traditional computer vision models process entire images in a single, global computation, a stark contrast to the human visual system's sequential, foveated glimpses. Recent research, inspired by length generalization issues in language models, reveals that this global processing in vision models leads to a critical failure in generalizing to tasks with increased complexity or length beyond their training distribution. The core finding is that strictly local, sequentia

At the World AI Conference in Shanghai, Xi Jinping announced 5,000 AI training slots for Global South countries and the launch of the "World Artificial Intelligence Cooperation Organization." Cooperation centers with ASEAN, the African Union, BRICS, and other alliances are planned to follow. China is systematically building a parallel AI governance structure outside Western influence. The article China's new World Artificial Intelligence Cooperation Organization is President Xi's clearest play y

Article URL: https://github.com/helasaoudi/llm-inspector Comments URL: https://news.ycombinator.com/item?id=48956776 Points: 4 # Comments: 1

<pre style='white-space:pre-wrap;width:81ex'>fix: custom Anthropic models work without maxTokens (#110617) * fix(agents): default Anthropic max output tokens * fix(agents): bound omitted Anthropic output cap</pre>

Plus: Russian cyberspies turn to infrastructure hacking, DHS repeatedly fails to realize it’d been hacked, a breach exposes an AI music generator’s scraping ways, and more.

The British AI Security Institute warns that open-weight models like GLM-5.2 and DeepSeek V4-Pro now trail closed frontier models in cyber capabilities by four to seven months. At the start of 2025, the gap was still six to ten months. It also found that safety measures on open models are largely ineffective, leaving defenders less time to prepare. The article Open-weight models now match frontier cyber performance from just four months ago at a fraction of the cost appeared first on The Decoder

link includes video with thinking captions, map state along with thinking of each player and replay file which can be viewed in game.chronodivide.com engine Comments URL: https://news.ycombinator.com/item?id=48956668 Points: 1 # Comments: 0

So, over-simplified, Oracle has agreed to build 7GW of data centres (costing about 300bn) in order to sell them to OpenAI who will pay 300bn over 5yrs. But if they don’t build the DC they don’t get cash, and if something something AI bubble then openAI cannot pay and so S&P500 just marked them down as one rung above junk (if they hit junk huge numbers of portfolios have to drop them etc etc) So basically it’s not likely but not impossible the Oracle can collapse under the weight of capex costs f

Article URL: https://www.youtube.com/watch?v=EwLW11Ucnps Comments URL: https://news.ycombinator.com/item?id=48956614 Points: 6 # Comments: 0

Now that Google has changed how its usage quotas are tallied, you might not get as many AI responses as you did before.

Hi all, I wanted to reshare this after some time. I've been working on evolving my service framework into an agent harness. As all of this AI tooling has evolved its become clear to me that agents, services and workflows are part of a core system of operations which would benefit from a single framework. Previously Go Micro only enabled service development. But now it includes Agents and Workflows. What does that mean? Well firstly services which are registered in a central registry automaticall

Github currently doesn’t let you unstar repositories in bulk. To resolve that let me introduce Workaround : a web app where users can not only unstar GitHub repos in bulk but also use AI filters and discover and star repos with AI. https://github.com/Ayush0054/workaround Comments URL: https://news.ycombinator.com/item?id=48956548 Points: 1 # Comments: 0

Article URL: https://www.theguardian.com/technology/2026/jul/16/google-workers-layoff-protections-ai Comments URL: https://news.ycombinator.com/item?id=48956537 Points: 6 # Comments: 0

Article URL: https://www.cnbc.com/2026/07/17/white-house-ai-access-anthropic-openai.html Comments URL: https://news.ycombinator.com/item?id=48956512 Points: 5 # Comments: 1

Hey everyone, I don’t use social media much. I spend around 30 minutes a day on X and sometimes Reddit. I stopped using Instagram a few months ago because my feed had become repetitive, distracting, and full of content I never intentionally asked for. And I usually have a purpose of scrolling, but this is getting harder day by day because of so much brainrot content. Whereas my partner is very found of scrolling, and I have noticed heat in our relationship when she take relationship advice from

<!-- SC_OFF --><div class="md"><p>I am more interested in ai ml but currently i am working as a graphic designer and i don't have a degree also except my 6 month diploma in graphic design.</p> <ol> <li><p>Any advice for how to get a job in ai ml , i started to learn maths and algorithms everyday evening but it looks like so much to learn ?</p></li> <li><p>Freshers where to start like data analyst or ml engineer or genai engineer ?</p></li> <li><p>where to contact employers because i don't have a

Today, I was asking Ai about the Canadian Wildfires and forest management. Instantly, the ai resorted to following the common narrative, and pushed macro human-caused climate change as an “objective fact.” I wanted to push back on the logic of this, because this narrative goes against core principles of scientific transparency. I wanted to highlight scientific principals: we are unable to run a double blind, controlled laboratory experiment on a duplicate of earth, that our historical proxy data

“Context bombing” tricks malicious AI agents into shutting down before they can do harm.

Article URL: https://owl.billpg.com/am-i-worried-ai-will-take-my-job/ Comments URL: https://news.ycombinator.com/item?id=48956304 Points: 2 # Comments: 0

Article URL: https://en.yna.co.kr/view/AEN20260716005600320 Comments URL: https://news.ycombinator.com/item?id=48956269 Points: 3 # Comments: 0

Article URL: https://fosnie.dev/ Comments URL: https://news.ycombinator.com/item?id=48956175 Points: 1 # Comments: 0

<p> 6.4x faster than llama.cpp, 3.9x faster than MLX </p> <p> <a href="https://www.producthunt.com/products/basert?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/1199683?app_id=339">Link</a> </p>

Article URL: https://ludic.mataroa.blog/blog/ai-mania-is-eviscerating-global-decision-making/ Comments URL: https://news.ycombinator.com/item?id=48956153 Points: 15 # Comments: 1