Intellectual hub of the topic
The transition from passive algorithms to autonomous systems capable of making decisions and acting in digital or physical environments presents new challenges for researchers. In this collection, we focus on methods for monitoring and verifying the behavior of AI agents. We analyze how to ensure the predictability of their actions, prevent them from exceeding their assigned mandates, and minimize the risks of unintended harm.
The Safetensors format has been officially adopted by the PyTorch Foundation – changing the approach to distributing model weights and securing agentic systems.
Salesforce AI Research explains how it is restructuring language model training for the agentic era – and why old approaches no longer work.
AI: Events
Security
Alibaba Cloud audited the security of 30,000 AI agent skills and discovered serious threats that require specialized protection tools.
AI: Events
Security
We'll explore one of the key threats to AI systems in business – prompt injection: how it works, why it's dangerous, and how to defend against it.
OpenAI is offering researchers rewards for finding ways to misuse AI – from attacks on agentic systems to data leaks through prompt manipulation.
We break down where and how malicious actors can attack AI models – from the training stage to production deployment.
Why the new competitive barrier in the world of AI isn't algorithms or data, but the ability to skillfully build agent management systems.
The startup Notch has shared how to create AI agents capable of performing real actions without losing control, even in regulated industries.
Nvidia has introduced OpenShell, a tool for safely running autonomous AI agents in a corporate environment. It isolates their actions and controls their access.
Want to dive deeper into the world
of neuro-creativity?
Be the first to learn about new books, articles, and AI experiments
on our Telegram channel!