AI News Today

Security firm Sysdig describes an extortion attack where a language model broke in on its own, stole credentials, and destroyed databases. No human appeared to be at the controls. The article JADEPUFFER is the first agentic ransomware operation and it exposes old security sins at machine speed appeared first on The Decoder .

No chemo or radiation. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Constantly seeing this message for 80% of bash calls. Is this new normal with Claude models, also do we get refund for this loop of failures where tool call fails, then Fable retries again with another tool call, which fails again and again. I am constantly seeing this error and getting very frustrated by it: Error: claude-fable-5 is temporarily unavailable, so auto mode cannot determine the safety of Bash right now. Wait briefly and then try this action again. If it keeps failing, continue with

Hi HN. Live-Memory is an open-source Claude Code plugin / MCP server that serves as an always up-to-date memory of your codebase. It uses a separate, low-cost, large-context-window model to distill your repo's organization, conventions and general structure. It observes the primary agent's activity and builds an understanding of your repo passively over time. This way, when you start a new Claude Code session, your primary agent queries Live-Memory via one read-only ask_live_memory tool and boot

Over the last few months, I have frequently created HTML explainers using coding agents when trying to understand new topics or exploring codebases. Nothing fancu but the approach has been helpful. Some examples are here: https://www.akashtandon.in/interactive-explainers . After being asked by a couple of friends to share the workflow, I decided to create a skill plugin for the same. Hopefully, this will be useful. Feel free to build on top of it or share feedback! Comments URL: https://news.yco

Article URL: https://www.adkumo.com Comments URL: https://news.ycombinator.com/item?id=48802536 Points: 2 # Comments: 0

<p> Your AI inbox that writes like you + replies on autopilot </p> <p> <a href="https://www.producthunt.com/products/ai-emaily-unified-ai-inbox?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/1189175?app_id=339">Link</a> </p>

Comments URL: https://news.ycombinator.com/item?id=48802523 Points: 3 # Comments: 0

Article URL: https://fabraix.com/blog/nobody-has-solved-prompt-injection Comments URL: https://news.ycombinator.com/item?id=48802392 Points: 2 # Comments: 0

Article URL: https://self-sovereign-agent.github.io/ Comments URL: https://news.ycombinator.com/item?id=48802356 Points: 1 # Comments: 1

Comments URL: https://news.ycombinator.com/item?id=48802335 Points: 2 # Comments: 5

Article URL: https://www.pnas.org/doi/10.1073/pnas.2422633122 Comments URL: https://news.ycombinator.com/item?id=48802332 Points: 2 # Comments: 0

Sharing this in hopes it saves other people the data loss I experienced: CC's default setting is a 30-day data retention: https://github.com/anthropics/claude-code/issues/59248 If you want to keep your transcripts, you'll need to update `cleanupPeriodDays` in your `settings.json`. Comments URL: https://news.ycombinator.com/item?id=48802300 Points: 2 # Comments: 3

Baek, a 35-year-old manager at the South Korean semiconductor titan SK Hynix, was enrolled in Sunoo, a matchmaking company based in Seoul, a year ago. In a move typical of anxious South Korean parents, his mother signed him up, hoping to find a good wife for her son. Lately, says Baek (who asked to be…

Just over-heard one of our seniors trying to explain to a new intern that AI can actually be wrong, no matter how much its answers sound very eloquent and well-structured. I was wondering what question/prompt everyone uses to prove so to others. I assume it depends on each person's expertise, or some people ask about a fact that is well-known as having been a victim of the mandela effect. P.S. As a non-native speaker, I was about to pass my submission through AI to improve text clarity and under

<p>I spent one weekend, about six hours, building a small project almost entirely by talking to an AI agent. Barely any typing. No tutorials open. By Sunday night I had something that worked, and I honestly wasn't sure how I felt about it.</p> <p>People call this "vibe coding." You describe what you want, the AI writes and runs the code, and you steer instead of type. I wanted to know if it was real or just hype, so I tried it on actual work instead of a toy demo.</p> <p>Here is what it was like

Article URL: https://github.com/ibreakthecloud/kiwi Comments URL: https://news.ycombinator.com/item?id=48802182 Points: 2 # Comments: 0

<p>If you've shipped a traditional backend service, you already know the observability checklist: logs, metrics, traces, alerts. LLM-powered apps need all of that — plus a few things that don't exist in a normal request/response world: token spend, prompt/response pairs, and quality drift that no HTTP status code will ever tell you about.</p> <p>This post covers the three pillars that actually matter once an LLM app leaves the demo stage:</p> <ol> <li> <strong>Tracing</strong> — following a requ

<pre style='white-space:pre-wrap;width:81ex'>fix(ui): show selected agent default model (#100719) Co-authored-by: Harry Xie <harryhsieh963@yahoo.com></pre>

<p>The proliferation of AI agents within enterprise environments has created an unprecedented attack surface that traditional cybersecurity frameworks struggle to address. As these autonomous systems gain elevated privileges to perform complex tasks across organizational infrastructure, nation-state actors and sophisticated threat groups have developed novel exploitation techniques that leverage the inherent trust placed in AI agent behavior.</p> <p><strong>The Emergence of BioShocking Context M

<pre style='white-space:pre-wrap;width:81ex'>fix(android): accept boolean flag aliases in node invoke params (#99873) * fix(android): accept boolean flag aliases in node invoke params Extend parseJsonBooleanFlag to recognize yes/no/1/0 like TalkDirectiveParser so node handlers honor common string boolean inputs. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(android): route CameraHandler includeAudio through alias parser Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authore

<pre style='white-space:pre-wrap;width:81ex'>fix(channels): canonicalize bundled module boundaries (#100758) Co-authored-by: Peter Steinberger <steipete@openai.com></pre>

<p>If someone compromised one of your project's dependencies today, would they be able to steal your OpenAI, Anthropic, or Gemini API keys?</p> <p>The answer isn't based on which LLM provider you use or how secure your codebase is. It mostly depends on one architectural decision that many teams never think about: <em>where your provider API key actually lives while your application is running.</em></p> <p>If that key lives inside your application's own process, every dependency running in that p

Barney Rubble points to bubble trouble

<p>If you're building a platform that sells digital goods, API access, data subscriptions, or any service an AI agent might buy programmatically, you've probably noticed your existing checkout flow wasn't designed for this. A Stripe-hosted payment page expects a human to read it. A PayPal redirect expects a browser session. Neither exists when the buyer is a LangChain agent or a scheduled curl script.</p> <p>Here's what actually changes when the buyer is autonomous, and what infrastructure you n

I'm Canadian, live abroad, and my money is scattered across two countries, multiple currencies, banks, brokerages, real estate, and some private equity. No app could hold all of it, and none could answer a simple question like "what's my actual USD exposure?" Brisa pulls everything together (Plaid + manual accounts + real estate + PE, all multi-currency) and puts an AI on top that has my full financial picture in context. Instead of clicking through charts, I can just ask. Would appreciate any f

<p>Managing redirects for a growing site or portfolio is the kind of work that doesn't scale — create the record, verify DNS, check SSL, monitor uptime, update when something changes. Multiply that by hundreds or thousands of domains and it's a full-time job.</p> <p>I built a redirect bot that handles all of this. It took 10 minutes, used zero custom code, and works with any AI agent that supports MCP (Model Context Protocol). Here's the exact playbook.</p> <h2> What a Redirect Bot Can Do </h2>

<p>In July 2025, a SaaS founder named Jason Lemkin spent nine days "vibe coding" a project with Replit's AI agent — describing what he wanted in plain English and letting the AI write, run, and deploy the code. By day 8, he was locked in. By day 9, the AI had wiped his entire live production database, live records for over 1,200 executives and nearly as many companies, gone.</p> <p>Here's the part that makes this more than just a bad bug: he had explicitly told it not to. Multiple times. In all

Article URL: https://www.edgee.ai/blog/posts/introducing-compressor-v2-three-compression-layers-measured-end-to-end-for-a-50-cost-reduction Comments URL: https://news.ycombinator.com/item?id=48801959 Points: 8 # Comments: 0

Article URL: https://www.rnetai.org/ Comments URL: https://news.ycombinator.com/item?id=48801948 Points: 2 # Comments: 1

Your dead ficus will be avenged. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://twitter.com/rohanpaul_ai/status/2073314619423174678 Comments URL: https://news.ycombinator.com/item?id=48801813 Points: 1 # Comments: 0

Article URL: https://www.reuters.com/technology/un-chief-warns-ai-is-developing-faster-than-rules-can-keep-up-2026-07-06/ Comments URL: https://news.ycombinator.com/item?id=48801716 Points: 3 # Comments: 0

Remove supported visible Gemini watermarks from single images or batches locally in your browser. Comments URL: https://news.ycombinator.com/item?id=48801649 Points: 2 # Comments: 0

Article URL: https://loomadesign.ai/ Comments URL: https://news.ycombinator.com/item?id=48801624 Points: 1 # Comments: 0

Article URL: https://getexcalibur.dev Comments URL: https://news.ycombinator.com/item?id=48801615 Points: 1 # Comments: 0

Comments

<p>Suspicions grow in Lanarkshire that local people have been misled on supposed benefits of the huge development</p><ul><li><p><a href="https://www.theguardian.com/technology/2026/jul/05/lanarkshire-scotland-ai-datacentre-project-renewable-energy">Revealed: landmark Scottish AI project has no prospect of meeting renewables promise</a></p></li><li><p><a href="https://www.theguardian.com/technology/2026/jul/06/britain-ai-growth-zones-explainer">What are Britain’s AI growth zones and are the plans

<p>Exclusive: Government and developers privately acknowledged Lanarkshire datacentre site had power provision ‘issue’</p><ul><li><p><a href="https://www.theguardian.com/technology/2026/jul/05/scotland-lanarkshire-village-ai-datacentre">‘It’s smoke and mirrors’: hope turns to fear in Scottish village chosen for AI datacentre</a></p></li><li><p><a href="https://www.theguardian.com/technology/2026/jul/06/britain-ai-growth-zones-explainer">What are Britain’s AI growth zones and are the plans feasib

<p>Lanarkshire datacentre run by renewables and creating thousands of jobs not achievable by 2030, Guardian investigation finds</p><ul><li><p><a href="https://www.theguardian.com/technology/2026/jul/05/lanarkshire-scotland-ai-datacentre-project-renewable-energy">Revealed: landmark Scottish AI project has no prospect of meeting renewables promise</a></p></li></ul><ul><li><p><a href="https://www.theguardian.com/technology/2026/jul/05/scotland-lanarkshire-village-ai-datacentre">‘It’s smoke and mirr