reinforcement learning
Harness-1 is a 20B-parameter retrieval subagent trained with reinforcement learning inside a stateful search harness built on gpt-oss-20b, designed to handle multi-step search and retrieval tasks more reliably than prompt-only agents.
AutoML
Startup Adaption has launched AutoScientist, a tool that automates the AI research and training process, letting models iteratively improve themselves with minimal human intervention.
Adobe
Adobe is integrating Claude Code-style agentic AI into Creative Cloud, enabling AI-driven creative workflows that could reshape how professionals produce and manipulate visual media at scale.
Open-Weight Models
Z.AI releases GLM-5.1, a 754B parameter open-weight model that achieves state-of-the-art results on SWE-Bench Pro and can sustain autonomous task execution for up to 8 hours.
agentic AI
As AI agents gain autonomy to execute code and access external systems, security becomes critical. These five architectural patterns help protect agentic AI from prompt injection, privilege escalation, and data leakage.
AI security
Security researchers demonstrate how hidden prompt injections in code repositories can hijack AI coding agents like Cline, exposing critical vulnerabilities in agentic AI systems.
agentic AI
New research proposes proxy state-based evaluation for multi-turn tool-calling LLM agents, addressing the challenge of scalable reward verification in complex agentic workflows.
Alibaba
Alibaba unveils Qwen3.5, positioning its latest AI model for the emerging era of autonomous AI agents with enhanced reasoning and task execution capabilities.
AI safety
New research examines how AI communities are splitting on human control approaches for autonomous agents, finding significant divergence in oversight philosophies that could shape the future of AI governance.
agentic AI
New research framework bridges traditional ML explainability methods with emerging agentic AI systems, proposing action-based interpretability for autonomous AI agents.
AI research
A new benchmark suite evaluates how well AI agents can perform frontier research tasks, measuring capabilities from literature review to hypothesis generation and experimental design.
agentic AI
A technical deep-dive into constructing enterprise-ready AI agents with hybrid retrieval systems, provenance tracking for citations, self-repair mechanisms, and persistent episodic memory.