AI Dynamics

Global AI News Aggregator

About

GENERATIVE AI

  • AI Agent Discovers Novel Adversarial Attacks Against LLMs
    AI Agent Discovers Novel Adversarial Attacks Against LLMs

    By

    @askalphaxiv

    –

    "Claudini: Autoresearch Discovers SoTA Adversarial Attack Algorithms for LLMs" This paper shows that an AI coding agent can autonomously invent jailbreak and prompt-injection attacks, and can even beat 30+ human designed methods. So on top of human red-teamers, the new LLM

    → View original post on X — @askalphaxiv

  • Foveated Diffusion: Efficient Spatially Adaptive Image Video Generation
    Foveated Diffusion: Efficient Spatially Adaptive Image Video Generation

    By

    @askalphaxiv

    –

    "Foveated Diffusion: Efficient Spatially Adaptive Image and Generation" This paper introduces the logic of human vision to diffusion models, where you generate full detail only when the viewer is looking, and becomes low detail in the periphery. With this setup, you can

    → View original post on X — @askalphaxiv

  • Day 83 GPU Programming: DeepSeek Multi-Head Latent Attention Optimization
    Day 83 GPU Programming: DeepSeek Multi-Head Latent Attention Optimization

    By

    @rasbt

    –

    Day 83/365 of GPU Programming Looking at DeepSeek's Multi-Head Latent Attention today. The last part of the AMD challenge series is to optimize an MLA decode kernel for MI355X where the absorbed Q and compressed KV cache are given and your task is to do the attention computation. A resource that really helped internalize what MLA does was @rasbt's incredible visual guide to attention variants in LLMs (luckily he posted that last week!), which covers everything from MHA to GQA to MLA to SWA, et cetera. If there's one place to get a visual intuition for recent attention mechanisms, it's this blog post. @jbhuang0604's video on MQA, GQA,MLA and DSA was the best conceptual intro I found on the topic and progressively builds up the ideas from first principles. The Welch Labs analysis of MLA is a great watch as well. Beautiful visualization of the changes DeepSeek made for MLA. Tried out a few kernels once I had a basic understanding of MLA and I think I'm slowly getting more comfortable with at least analyzing kernels. levi (@levidiamode) Day 82/365 of GPU Programming Taking a closer look at Mixture of Experts today, so I can write better MoE kernels. Specifically, to optimize an MXFP4 MoE fused kernel for the GPU Mode challenge. I haven't had much prior exposure to MoEs, so lots of new concepts I learned today. Luckily I found the best intro to MoEs thanks to @MaartenGr visual overview of the topic. I then watched @tatsu_hashimoto's amazing Stanford CS336 lecture on MoEs, which added deeper context around why MoEs are gaining popularity, FLOPs, OLMoE, infra complexity, routing functions (mindblown this works so well…), expert sizes, training objectives, top k routing and DeepSeek variations. Once I had a basic understanding I started playing around with the some AITER kernels but progress there is tbd. Also had a nice chat with @juscallmevyom (who was kind enough to reach out!) about the AMD kernels and the challenge of materialization overhead. — https://nitter.net/levidiamode/status/2037297869518950430#m

    → View original post on X — @rasbt, 2026-03-27 22:49 UTC

  • Democratizing AI: 50K Models and Local Inference Freedom

    By

    @clementdelangue

    –

    Next steps:
    – enable the 50,000 models available in inference providers
    – enable the 3,000,000 models available on HF
    – local free fast inference with llama.cpp
    – train and bring your own model! We don't want a world where you're forced to choose between two or three lookalike

    → View original post on X — @clementdelangue

  • Hugging Face Integration Added to Hermes Agent Platform
    Hugging Face Integration Added to Hermes Agent Platform

    By

    @julien_c

    –

    We have integrated @huggingface as a first-class inference provider in Hermes Agent. When you select Hugging Face in the model picker it now shows 28 curated models organized by use case, with a custom option for the 100+ other models they serve. clem 🤗 (@ClementDelangue) Been really cool to see the traction of @NousResearch Hermes Agent, the open source agent that grows with you! Hermes Agent is open-source and remembers what it learns and gets more capable over time, with a multi-level memory system and persistent dedicated machine access. Starting today, you can use a bunch of @huggingface open-source models thanks to our inference provider partners. Let's go open agents! — https://nitter.net/ClementDelangue/status/2037634211973140898#m

    → View original post on X — @julien_c, 2026-03-27 22:15 UTC

  • AI Voice Support: Game-Changer for Customer Service

    By

    @datachaz

    –

    Gobsmacked by the skepticism in the comments. We’ve suffered 30 years of awful automated menus and slow, unhelpful chat reps… basically since phones existed. Instant, helpful AI voice support will be a massive win-win in many (not all!) cases. Saying 'it’s not ready' ignores

    → View original post on X — @datachaz

  • Google Turbo Quant LLM Running Locally on MacBook Air

    By

    @kimmonismus

    –

    Google Turbo Quant running Locally in Atomic Chat MacBook Air M4 16 GB
    Model: QWEN3.5-9B
    Context window: 100000 Summarising 50000 words in just seconds..
    You can do 3x larger context window, processing 3x faster than before! They are first that have integrated Google

    → View original post on X — @kimmonismus

  • Major AI Tools Coming Soon from OpenAI Google Anthropic
    Major AI Tools Coming Soon from OpenAI Google Anthropic

    By

    @kimmonismus

    –

    Google is also working on a new internal ai tool called „agent smith“. It seems very popular within the company. So we are waiting for – OpenAI „spud“ (few weeks)
    – Claude Mythos (Q3/2026)
    – Googles Agent Smith (later this year)

    → View original post on X — @kimmonismus

  • AI Automation Eliminating Code Reviews and Developer Workflows

    By

    @latentspacepod

    –

    nice followup from Ankit: https://
    x.com/ankitxg/status
    /2036487355981234377?s=20
    … and @TheNoahHein : https://
    x.com/TheNoahHein/st
    atus/2037573208707137639
    … and https://
    molochinations.substack.com/p/no-more-code
    -reviews-lights-out-codebases?triedRedirect=true
    … it is coming!

    → View original post on X — @latentspacepod

  • Teaching AI to Create Beautiful Data Visualizations Autonomously
    Teaching AI to Create Beautiful Data Visualizations Autonomously

    By

    @randal_olson

    –

    this is a great goal for AI assistance. It is so much work to make nice visuals and also conceptually hard. would be amazing to help people embrace dataviz do's and get past the don'ts. Randy Olson (@randal_olson) What happens when you teach an AI agent to make beautiful dataviz? New project: I’m building a quality harness for my AI to make better dataviz. One example attached – it researched, made this, and wrote the article autonomously. Follow the project here: randalolson.com/beautiful-ch… — https://nitter.net/randal_olson/status/2037551085741891731#m

    → View original post on X — @randal_olson, 2026-03-27 21:19 UTC