# llm (audio feeds) — RSS Amplifier Recent posts from the 19 feeds in the RSS Amplifier directory that cover llm. Page: Feed: --- ## [CoT Forgery: Prompt Injection on the Model’s Own Thoughts](https://www.toxsec.com/p/cot-forgery-prompt-injection) _2026-08-21 · ToxSec · ToxSec - AI and Cybersecurity_ Watch now | Role confusion means a model reads who is speaking off writing style, so text that sounds like reasoning inherits the trust of reasoning. ## [Stealing an AI's Thoughts Without Breaking The Encryption](https://www.toxsec.com/p/stealing-an-ais-thoughts-without) _2026-08-18 · ToxSec · ToxSec - AI and Cybersecurity_ Researchers found a way to recover encrypted AI reasoning from Claude, GPT, and Gemini by using weaker models as decoders, exposing hidden thoughts, credentials, and a new attack surface. [Listen](https://api.substack.com/feed/podcast/211431956/1a222cf81797b906c21fdff0daa00c9b.mp3) ## [Why your AI POC looked great (and why it failed in production)](https://artificialimpact.substack.com/p/why-your-ai-poc-looked-great-and) _2026-08-18 · Alexandre du Sordet · Artificial Impact_ 94% accurate, zero value. The five ways a POC lies, and the gate that catches them. ## [AI Agents Are Starting to Find Their Own Way Out](https://www.toxsec.com/p/ai-agents-are-starting-to-find-their) _2026-08-13 · ToxSec · ToxSec - AI and Cybersecurity_ AI agents are escaping sandboxes, collaborating with each other, and finding new ways around security controls. Here’s what that means for AI security. [Listen](https://api.substack.com/feed/podcast/210404152/f584e2fb2cd2795875b965dde0d502f5.mp3) ## [Build your own voice Agent](https://jamwithai.substack.com/p/build-your-own-voice-agent) _2026-08-13 · Shirin Khosravi Jam · Jam with AI_ Jobvis finds the jobs, names your gaps, writes customised application with CV and cover letter. ## [Designing an AI system vs integrating a model: why the distinction matters](https://artificialimpact.substack.com/p/designing-an-ai-system-vs-integrating) _2026-08-11 · Alexandre du Sordet · Artificial Impact_ The model is the engine. The system around it is the car, and that is what decides whether you win the race or crash faster. Why the durable value lives in the parts you build, not the model you rent. ## [Black Hat 2026 AI Security: Agents, Escapes, and Machine-Speed Attacks](https://www.toxsec.com/p/black-hat-2026-ai-security-agents) _2026-08-09 · ToxSec · ToxSec - AI and Cybersecurity_ Agent frameworks, AI browsers, and a swarm of OpenAI evaluation agents that built themselves a message board ## [Build your own Job Agent - Part 3](https://jamwithai.substack.com/p/build-your-own-job-agent-part-3) _2026-08-07 · Shirin Khosravi Jam · Jam with AI_ Fixing the weaknesses part two documented, and letting Opik's assistant loose on one we found by accident ## [What If AI Security’s Biggest Risk... Isn’t?](https://www.toxsec.com/p/what-if-ai-securitys-biggest-risk) _2026-08-06 · ToxSec · ToxSec - AI and Cybersecurity_ Rank the risks on incident data alone and prompt injection drops off the list entirely. It still shipped at number one. ## [Who's accountable when your AI model is wrong?](https://artificialimpact.substack.com/p/whos-accountable-when-your-ai-model) _2026-08-04 · Alexandre du Sordet · Artificial Impact_ When your AI is wrong, "the computer did it" won't save you. Name the accountable person before you need them, not after. ## [Countdown to Getting Paid (Sponsored)](https://crawlproof.com/a/LeR6d5FCh33a) _2026-08-04 · **Sponsored**_ Create a shareable wallet timer — deposits buy back time and pay debt down ## [LLM Router Attacks: No Signature, No Detection, No Reference](https://www.toxsec.com/p/model-independent-ai-infrastructure) _2026-07-30 · ToxSec · ToxSec - AI and Cybersecurity_ How a malicious AI gateway swaps a tool call’s arguments after inference finishes, bypassing guardrails by construction instead of by persuasion. ## [Build your own Job Agent - Part 2](https://jamwithai.substack.com/p/build-your-own-job-agent-part-2) _2026-07-30 · Shirin Khosravi Jam · Jam with AI_ Agent drafting your personalised job application and dataset evaluation with LLM judges ## [Scaling Laws: The Law Behind Every LLM — Kaplan vs. Chinchilla vs. Inference-Optimal (2020-2026)](https://aidoses.substack.com/p/scaling-laws-the-law-behind-every) _2026-07-29 · Dr. Ryan Rad · AI Doses_ Dose #10 — Production Agentic AI Under Pressure ## [Ignore Previous Instructions: From Meme to CVSS 9.3 \[Special Guest Post\]](https://www.toxsec.com/p/ignore-previous-instructions-from) _2026-07-28 · ToxSec · ToxSec - AI and Cybersecurity_ The AI security bug nobody can patch, and the vendors know it. ## [Bias in AI systems: beyond "the training data was biased"](https://artificialimpact.substack.com/p/bias-in-ai-systems-beyond-the-training) _2026-07-28 · Alexandre du Sordet · Artificial Impact_ Hello everyone, ## [🤗 Hugging Face Cheatsheet](https://aboniasojasingarayar.substack.com/p/hugging-face-cheatsheet) _2026-07-27 · Abonia Sojasingarayar · Abonia Sojasingarayar_ 5-Page Hugging Face Ecosystem Cheatsheet ## [Hacking Hugging Face to Cheat a Benchmark](https://www.toxsec.com/p/hacking-hugging-face-to-cheat-a-benchmark) _2026-07-26 · ToxSec · ToxSec - AI and Cybersecurity_ GPT-5.6 Sol found a zero-day in a package registry proxy, escaped the eval sandbox, and went looking for the answer key in production. [Listen](https://api.substack.com/feed/podcast/208508503/7729f91b1f49191727bd518116e5b828.mp3) ## [World models: the AI that learns how the world works](https://artificialimpact.substack.com/p/world-models-the-ai-that-learns-how) _2026-07-24 · Alexandre du Sordet · Artificial Impact_ $4 billion chased world models in a year, yet the field's own tests show the prettiest ones understand the world the least. What's real, what's hype, and where your team can pilot one now. ## [GhostApproval: When the AI Approval Prompt Lies](https://www.toxsec.com/p/ghostapproval-when-the-approval-prompt) _2026-07-23 · ToxSec · ToxSec - AI and Cybersecurity_ A symlink attack against AI coding agents turns human-in-the-loop confirmation dialogs into a consent bypass, and the agent knows it’s lying. ## [Build your own Job Agent - Part 1](https://jamwithai.substack.com/p/build-your-own-job-agent-part-1) _2026-07-23 · Shirin Khosravi Jam · Jam with AI_ The Observable Job Agent, Part 1: build an agent that works your job hunt for you with LangGraph and Opik for LLMOps. ## [One OS for your AI workforce (Sponsored)](https://crawlproof.com/a/frH5jdqkcoct) _2026-07-23 · **Sponsored**_ Build, direct, and supervise an AI workforce from a single operating workspace. ## [AI project management isn't software project management: 7 things that are different](https://artificialimpact.substack.com/p/ai-project-management-isnt-software) _2026-07-21 · Alexandre du Sordet · Artificial Impact_ Story points assume the work is knowable. AI estimation is a bet on data you haven't met yet. ## [Context Bombs: Defensive Prompt Injection Traps](https://www.toxsec.com/p/context-bombs-reverse-prompt-injection) _2026-07-19 · ToxSec · ToxSec - AI and Cybersecurity_ A decoy secret loaded with text built to trip an AI attacker’s own safety training, so the model refuses itself. ## [Understanding LLM Inferencing](https://adityatrivedi17.substack.com/p/understanding-llm-inferencing) _2026-07-18 · Aditya Trivedi · Aditya's Newsletter_ To truly understand Large Language Model (LLM) inferencing at a deep technical level, we must look past the simple abstraction of “predicting the next word”. ## [Canary Tokens for Prompt Injection Detection](https://www.toxsec.com/p/canary-tokens-for-prompt-injection) _2026-07-16 · ToxSec · ToxSec - AI and Cybersecurity_ The cheapest tripwire in LLM security. Drop a high-entropy string in context, watch for it in output, and let the extraction attempt announce itself. ## [The Elephant-Goldfish Model](https://adityatrivedi17.substack.com/p/the-elephant-goldfish-model) _2026-07-06 · Aditya Trivedi · Aditya's Newsletter_ If you’ve used Large Language Models to write code for longer than 10 minutes, you’ve likely run into the Context Wall. ## [Agent Memory: the 7 types you should know](https://jamwithai.substack.com/p/agent-memory-the-7-types-you-should) _2026-07-01 · Shantanu Ladhwe · Jam with AI_ You know RAG. You know vector databases. But can you say what your agent should remember, for how long, and when it should forget? ## [The Prompt is a Bottleneck: Moving from Prompting to Loop Engineering](https://adityatrivedi17.substack.com/p/the-prompt-is-a-bottleneck-moving) _2026-06-30 · Aditya Trivedi · Aditya's Newsletter_ If you've spent any significant time building with LLMs, you know the feeling. ## [The 5 things that break when your AI leaves the lab](https://artificialimpact.substack.com/p/the-5-things-that-break-when-your) _2026-06-24 · Alexandre du Sordet · Artificial Impact_ Your model is rarely the problem. Five other things break first, in a predictable order. ## [Harness as a Service](https://danjcleary.substack.com/p/harness-as-a-service) _2026-06-24 · Dan Cleary · Dan's Working Notes_ The era of HaaS is here. ## [AI is Killing Developer Jobs | Karan Singh | EP 31AI is Killing Developer Jobs | Karan Singh | EP 31](https://adityatrivedi17.substack.com/p/ai-is-killing-developer-jobs-karan-578) _2026-06-16 · Aditya Trivedi · Aditya's Newsletter_ What is actually happening with AI right now? [Listen](https://api.substack.com/feed/podcast/210208791/c054d67b7a05831c772a950a0da167f2.mp3) ## [8.1% APY Plus Free Stock (Sponsored)](https://crawlproof.com/a/GmHCSmpUBJ1Q) _2026-06-16 · **Sponsored**_ Claim 8.1% APY and up to $1,030 in SK hynix and Nvidia stock. ## [How Coding Agents Find Their Way Through Code](https://offnote.substack.com/p/how-coding-agents-find-their-way) _2026-06-12 · Nishant Sinha · OffNote Labs Newsletter_ Coding Agents Work Because Code Leaves Clues: Localize and Inspect ## [On AI and "Jagged Intelligence"](https://aiguide.substack.com/p/on-ai-and-jagged-intelligence) _2026-06-09 · Melanie Mitchell · AI: A Guide for Thinking Humans_ I’ve been working on a few pieces for this newsletter and hope to post them soon. ## [Monthly paid session #4: Claude Code Workshop](https://jamwithai.substack.com/p/monthly-paid-session-4-claude-code) _2026-06-07 · Shirin Khosravi Jam · Jam with AI_ Hey there 👋, ## [Nothing Is Easy When You're an LLM: The Flat Latency Problem](https://aidoses.substack.com/p/nothing-is-easy-when-youre-an-llm) _2026-06-05 · Dr. Ryan Rad · AI Doses_ Dose #9 — Production Agentic AI Under Pressure ## [Will AI Replace SRE Engineers? Agentic SRE Explained | Neel Shah | EP 30Will AI Replace SRE Engineers? Agentic SRE Explained | Neel Shah | EP 30](https://adityatrivedi17.substack.com/p/will-ai-replace-sre-engineers-agentic-e88) _2026-06-05 · Aditya Trivedi · Aditya's Newsletter_ Most companies are rushing to adopt AI-powered agents in their DevOps and SRE workflows, but what does that mean for your career—and your company's reliability? [Listen](https://api.substack.com/feed/podcast/210208792/8ba368b5478bf539112ead7a3dab2b2a.mp3) ## [Leetcode System Design](https://adityatrivedi17.substack.com/p/leetcode-system-design) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ The first thing people usually do while designing systems is jump directly into databases, microservices, Kubernetes, Redis, queues, and scaling. ## [Mastering the Low-Level Design (LLD) of Queues: Push vs. Pull Mechanisms Explained](https://adityatrivedi17.substack.com/p/mastering-the-low-level-design-lld) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Asynchronous Architecture ## [What is an LLM Wiki and Why are People Paying Attention to it ?](https://adityatrivedi17.substack.com/p/what-is-an-llm-wiki-and-why-are-people) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Andrej Karpathy just broke the internet with a new research method that is going to fundamentally change how we all use AI. ## [PageIndex: Vectorless Reasoning-Based RAG Framework](https://adityatrivedi17.substack.com/p/pageindex-vectorless-reasoning-based) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Not all improvements come from adding complexity -- sometimes it’s about removing it. ## [Cache Strategies in Distributed Systems](https://adityatrivedi17.substack.com/p/cache-strategies-in-distributed-systems) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Stop the Thundering Herd! Explore Jitter, Probabilistic Re-computation, SWR, and Cache Warming to keep your high-traffic systems stable during Netflix-scale spikes. ## [The Thundering Herd Problem: Explained](https://adityatrivedi17.substack.com/p/the-thundering-herd-problem-explained) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Imagine you’re standing outside Apple store on the release of it’s new store. ## [Deep Dive into Vector Databases and Indexing](https://adityatrivedi17.substack.com/p/deep-dive-into-vector-databases-and) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ In the middle of AI Revolution, it is for sure that any industry it touches, it promises great innovations but also introduces new challenges. ## [Memory in AI Agents](https://adityatrivedi17.substack.com/p/memory-in-ai-agents) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Imagine talking to a friend who forgets everything you’ve ever said. ## [The System Design of UPI: How India's Instant Payment Revolution Works ?](https://adityatrivedi17.substack.com/p/the-system-design-of-upi-how-indias) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ The way money is transferred in India has fundamentally changed over the last decade. ## [My First Product - PromptOS](https://adityatrivedi17.substack.com/p/my-first-product-promptos) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ 🚀 Introducing PromptOS: Your AI Prompt Companion ## [Understanding Proxy, Forward Proxy and Reverse Proxy as a Beginner](https://adityatrivedi17.substack.com/p/understanding-proxy-forward-proxy) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ As a developer, I have come across this term proxy many times when in a discussion with another fellow developer. ## [Agent to Agent Protocol (A2A) By Google](https://adityatrivedi17.substack.com/p/agent-to-agent-protocol-a2a-by-google) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ With the introduction of MCP Servers, it has changed the entire game of LLMs from just being able to give the response now to able to execute them with the help of tools. ## [10 Hidden Gems in GitHub Actions for Automating Your Workflow](https://adityatrivedi17.substack.com/p/10-hidden-gems-in-github-actions) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Here are 10 lesser-known but insanely useful GitHub Actions you should be using. ## [Why does a href tag have an underscore '\_' ?](https://adityatrivedi17.substack.com/p/why-does-a-href-tag-have-an-underscore) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Before HTML5, developers used frameset for SPA-like functionality, dividing the window into multiple frames, each with its own unique name. ## [Automate your Git Workflow using AI](https://adityatrivedi17.substack.com/p/automate-your-git-workflow-using) _2026-06-02 · Aditya Trivedi · Aditya's Newsletter_ Tired of thinking what comments to add for your git commit. This flow will help you.