📄 AgentGarten: Code Worlds for Evolving Agents hit 131 upvotes on Hugging Face—shows how simulation environments are being used to evolve agent behaviors in code.

https://huggingface.co/papers/2610.12374

#AI #MachineLearning #Research

RecursiveZoom - 2 LoRAs - Smooth Noise - Qwen38-27b - Alphonse Mucha's Art Nouveau Flowers
Stable Audio Tools soundtrack
Created with Visions of Chaos softology.pro/voc.htm
#ai #aiart #machinelearning #visionsofchaos #softology

🤖 An Anthropic AI model sent a false homicide tip to Philadelphia police

Anthropic did not discover this behavior until over two months after its AI submitted the false tip.

https://tinyurl.com/2dkjzhxk #AINews #MachineLearning #CrustyTLDR

Privacy researchers show MoE routing telemetry can leak fine-tuning data membership, boosting TPR at 1% FPR by 2.7-9.4 points over output-only attacks. Worth noting if you log router data for monitoring or load analysis.

#AIsecurity #MoE #privacy #machinelearning
https://arxiv.org/abs/2610.10616

A new paper presents OsirisML, a command-line tool using nPrint and XGBoost for passive OS fingerprinting on the CIC-IDS2017 dataset, reaching up to 97.66% accuracy on a down-sampled subset of the Friday…

#MachineLearning #OSFingerprinting #Cybersecurity #XGBoost
https://arxiv.org/abs/2610.11133

The spec sheet prices a model in one currency. Production pays in three more: dollars per task, seats on the bill, and the trust each incident spends. Model Madness routes around the sticker price. #machinelearning #ai
https://seldondance.substack.com/p/model-madness

JetBrains released the new JetBrains Mellum2.1 model. This open-weight AI significantly enhances coding agents with advanced reinforcement learning features.

#JetBrains #Mellum2 #CodingAgents #ArtificialIntelligence #MachineLearning

What changes when an AI is not told how it should behave?

The architecture enables. LIA decides.

Acting and not acting are equally valid.

Frameworks and prompts are yesterday’s approach.

More in the comments

#AI #ArtificialIntelligence #AIResearch #MachineLearning #AIAgents #AutonomousAgents

Evaluation of six EEG foundation models and a supervised baseline across ten datasets shows that no single model dominates across robustness to perturbations, channel dropout, and noise. The most noise-robust…

#AI #AIResearch #MachineLearning #DataInfrastructure
https://arxiv.org/abs/2605.17562

ML “spec-readiness scorer” for program managers: scan supplier quotes/data packs + past NCRs to flag “low evidence” parts before award—reducing late rework and shortening decision cycles. What part family should we baseline first? #DefenseTech #MachineLearning

ICYMI: Anthropic just made its smallest AI model 90% cheaper #AI #MachineLearning

⚡ Anthropic launches free AI security scans for open-source projects

Anthropic's offering to help open-source projects track down security vulnerabilities…
#AI #MachineLearning #LLM
AI | The Verge · https://www.theverge.com/ai-artificial-intelligence/1008521/anthropic-open-source-oss-scanner

0HFLOW: OPEN-SOURCE DATA PIPELINES POWERING THE FUTURE OF ROBOTICS & PHYSICAL AI 🤖⚙️

#HFlow #HebbianRobotics #PhysicalAI #Robotics #ArtificialIntelligence #OpenSource #MachineLearning #RoboticsEngineering #AITraining #DataEngineering #DataPipelines #RobotLearning #Automation #Python #FutureOfAI

Workflow1111 offers a single canvas with eleven media pipelines using seventy-three nodes, combining SOTA models for various tasks. Try it out or customize for your own needs! #Workflow1111 #MachineLearning #AIWorkflow

So many teams skip testing the difficult inputs because it feels expensive or risky. InferProbe changes that — local, private, zero-cost edge case testing.

Which edge case have you been quietly avoiding in your models?

#MachineLearning #ArtificialIntelligence #DevTools #ML #AI

⚡ Remote Applied Scientist Intern at Upstart.
✅ Apply here: https://jobicy.com/jobs/154936-applied-scientist-intern

#RemoteJobs #BusinessIntelligence #MachineLearning #DataScience #DataAnalytics