Search results

According to the tag data, the following results have been found:

Protection Without Encryption: Paradox or Alternative?

Is it possible to protect data without encryption? In 2025, as algorithms become outdated, quantum computing is on the rise, and most data breaches are caused by human error, this question turns from trivial to strategic. The article covers alternative approaches, unexpected solutions, real-world project cases, practical code examples, and personal experience implementing data protection where encryption is either inappropriate or insufficient. Things will get unconventional, sometimes bordering on hacks, but always to the point.

Digest of Information Security Regulation. January – March 2025

Hello, tekkix! We continue our series of reviews on laws, orders, decrees, and regulatory initiatives related to information security. In this article, we cover what has changed in information security regulation since the beginning of 2025.

Personal AI assistant on your data. Part 1: Vector Database ChromaDB + DeepSeek | GPT

Today we’ll talk about a topic that sparks lively interest among many developers and AI enthusiasts — integrating large language models like DeepSeek or ChatGPT with your own knowledge base.

Restoring a Damaged Linux via chroot

Have you ever had to open up a Linux machine that wouldn't boot, even after ensuring the BIOS settings were correct and confirming there were no serious hardware errors in the system?

How to Protect Wealth Using IT and "Hostile" Interfaces?

A good interface should help the user. But what if I say that sometimes a good interface should hinder them?

In pursuit of the unknown: how an ML model learned to detect malware

Hello everyone! This is Ksenia Naumova. At Positive Technologies, I research malicious network traffic and improve tools for analyzing it at the security expert center. Recently, we were tasked with creating an ML model to detect malware in the network. It had to recognize not only the malware we had previously detected, but also new threats that emerge in large numbers every day. As a first experiment, we decided to build a model to work with traffic transmitted via the HTTP protocol, as our products successfully decrypt TLS sessions, which often contain a lot of interesting data. In this article, I will describe in detail how we trained the model and share information about the mistakes we made.

How we revived DPED: An AI project to enhance mobile photos to DSLR-quality images

Hello, tekkix! The team of the regional scientific and educational center "Artificial Intelligence and Big Data Analysis" at NSTU named after R. E. Alekseev is here.

We hacked Google Gemini and downloaded its source code

In 2024, we released a post We Hacked Google A.I. for 50,000, which described how our group consisting of Roni “Lupin” Carta, Joseph “rez0” Tacker, and Justin “Rhynorater” Gardner went to Las Vegas in 2023 to search for vulnerabilities in Gemini at the Google event LLM bugSWAT. This year, we repeated our trip…

Special architecture for data centers that handle AI operations: how Rubin surpasses Blackwell

Recently, NVIDIA announced that the release of the GB200 NVL72 supercomputer on the Blackwell architecture is delayed, and the GB300 superserver will be released in mid-2025. It turns out that these devices will be released almost simultaneously with the Rubin architecture, although the latter is positioned as next-gen. In the article, we will consider these technologies and try to understand why the company is in such a hurry to release the new platform.

Making macOS safer

This is a guide about security, not privacy.

∇²DFT — a new dataset and benchmark for solving quantum chemistry problems using neural networks

My name is Kuzma Khrabrov, I am a research engineer at AIRI and I work on tasks at the intersection of machine learning, quantum chemistry, and computational biology. Together with the team, we create new datasets, train new models, and come up with methods to solve both fundamental and practical problems.

How OSINT reunited two long-lost soldiers

Faces and names mentioned in this blog are real and used with their permission. Some details have been edited or hidden to protect the privacy of others.

How LLMs are changing system architecture: from simple data pipelines to intelligent autonomous agents

At every technical conference lately, the word "agents" is sure to be mentioned. They are presented in different ways: as the next step after RAG, as a silver bullet for solving all problems, and as an absolute replacement for all classic pipelines. And those who do not use agents yet are hopelessly behind progress.

Worse than a flood, scarier than a fire: how to prepare your backups for a ransomware virus visit

For decades, backups have primarily protected us from physical equipment failure and accidental data corruption. A good backup system should survive a fire, a flood, and then quickly enable the business to continue normal operations. But another problem has emerged, which is much more likely than a flood and from which fireproof barriers and physical separation of sites in different cities do not save.

OSINT. Now what is the buzz about?

We at IDX, a company engaged in legal personal data verification, are naturally interested in everything related to PD, even if it goes beyond our operational activities. At the same time, we naturally keep one thought in mind - is it time for us to get involved in this too, to complement and enrich our services.

What determines the security of a quantum network? Part 1

Quantum key distribution [1] is one of the most rapidly developing areas in modern science. The most important advantage of QKD is the security of personal data transmission with quantum-distributed keys, based on the laws of quantum physics, rather than mathematical algorithms. The latter can be hacked, but it is impossible to deceive the fundamental laws of physics.

8 best tips for outsourcing data annotation

Any CV project starts with annotating large volumes of images and videos. Only successful results and high-quality data ensure that the model can be trained correctly.