about.subtitle
This collection brings together research and analytical materials focused on the resilience of systems against external threats and internal structural failures. We view security not as a set of constraints, but as a fundamental prerequisite for technological advancement. Central to our focus is the problem of alignment, where it is critically important that complex algorithms operate within the framework of defined human values and do not lead to unpredictable consequences.
AI: Events
Security
OpenAI has discovered that modern AI models struggle to control their own thought processes – and this could be a crucial defense against manipulation.
OpenAI has released the system card for GPT-5.4 Thinking – a document explaining the new model's working principles, limitations, and safety level.
OpenAI introduced GPT-5.3 Instant – a lightweight model focused on speed and accessibility, balancing performance and safety.
A mathematical model for dengue fever that incorporates age and vaccination serves as a tool for predicting the disease's long-term behavior.
AI: Events
Technical context • Security
We're breaking down how MCP server and client security works and why properly configured access control is crucial for any agent-based system.
LightOn has released EDiTh, an open-source benchmark that allows testing corporate search on realistic documents without the risk of leaking confidential data.
Want to know about new
experiments first?
Subscribe to our Telegram channel — we share all the latest
and exciting updates from NeuraBooks.