All news with #how to tag

🧬 This guide demonstrates step-by-step fine-tuning of MedGemma (a Gemma 3 variant) to classify breast histopathology images using the public BreakHis dataset and a notebook-based workflow. It highlights practical choices—using an NVIDIA A100 40 GB, switching from FP16 to BF16 to avoid numerical overflows, and employing LoRA adapters for efficient training. The tutorial reports dramatic accuracy gains after merging LoRA adapters and points readers to runnable notebooks for reproducibility.

🔒 Google Cloud's Private NAT enables secure private-to-private translation to connect networks with overlapping or non-routable IPv4 ranges without running NAT appliances. As a managed Cloud NAT feature, it delivers high availability, automatic scalability, and centralized control for hybrid and multi‑VPC topologies. The post includes practical gcloud examples and Network Connectivity Center use cases to guide implementation.

🔒 This article by Stan Kaminsky reviews Athanasios Giatsos’ Security Analyst Summit 2025 talk and explains why malicious browser extensions are a major blind spot for organizations. It outlines how extensions can access cookies, local storage, proxy settings, clipboard and screen capture, enabling session and account theft, espionage, ad fraud and crypto theft, and why Manifest V3 reduces but does not eliminate risk. Practical controls described include formal extension policies and allowlists, disabling developer mode, version pinning and testing of updates, EDR and SIEM-based monitoring, and the use of specialized vetting tools for deeper analysis.

🚀 Google Cloud has launched the Production-Ready AI Learning Path, a free curriculum designed to guide developers from prototype to production. Drawing on an internal playbook, the series pairs Gemini models with production-grade tools like Vertex AI, Google Kubernetes Engine, and Cloud Run. Modules cover LLM app development, open model deployment, agent building, security, RAG, evaluation, and fine-tuning. New modules will be added weekly through mid-December.

🚀 Google Cloud introduces hands-on labs for Gemma 3, a family of lightweight open models offering multimodal (text and image) capabilities and efficient performance on smaller hardware footprints. The labs present two deployment paths: a serverless approach using Cloud Run with GPU support, and a platform approach using GKE for scalable production environments. Choose Cloud Run for simplicity and cost-efficiency or GKE Autopilot for control and robust orchestration to move models from local testing to production.

🔒 Implementation and certification of ISO standards or an ISMS frequently falter due to avoidable organizational and technical mistakes. The article outlines nine recurring issues — from weak management sponsorship and treating certification as a one‑off task to poor employee engagement, inadequate skills development, dishonest assessments, and insufficient follow‑up. For each pitfall it recommends practical remedies such as executive commitment, clear planning, targeted training, honest risk analysis, automation where appropriate, and adequate resourcing to make the management system functional and sustainable.

🧭 This post explains when to use sub-agents versus packaging agents as tools when building multi-agent systems with Google's Agent Development Kit (ADK). It contrasts agents-as-tools — encapsulated, stateless specialists invoked like deterministic function calls — with sub-agents, which are stateful, context-aware delegates that manage multi-step workflows. The guidance highlights trade-offs across task complexity, context sharing, reusability, and autonomy, and illustrates the patterns with data-agent and travel-planner examples to help architects choose efficient, scalable designs.

🔒 Most email blackmail attempts are mass scams that exploit leaked personal data and fear to extort cryptocurrency from victims. The article outlines common themes — fake device hacks, sextortion, and even fabricated death threats — and describes regional campaigns where attackers impersonate law enforcement in Europe and CIS states. It highlights detection signs and practical defenses, urging verification, use of reliable security solutions, and reporting threats through official channels.

🚀 This codelab walks developers through building a functional travel chatbot using Google's Gemini via the Vertex AI SDK. It explains how to connect a web frontend to Gemini, craft system instructions to shape assistant behavior, and enable function-calling to fetch live data such as geocoding and weather. No advanced ML expertise is required; the lab provides step-by-step code samples, API usage, and practical recommendations for iterating prompts so you can produce a working, production-ready demo.

🔁 This AWS Security Blog post by Samuel Folkes outlines a practical approach to migrating authorization from Open Policy Agent (OPA) and Rego to Amazon Verified Permissions using the Cedar policy language. It highlights key benefits: a fully managed service, reduced operational overhead, and significant performance gains. The article walks through schema design, common translation patterns (RBAC, ABAC, ReBAC), application integration changes, testing practices, and a phased deployment strategy to compare and validate behavior during migration.

🔒 A growing scam exploits WhatsApp’s screen-sharing feature to trick users into exposing verification codes, passwords and banking details during video calls. Attackers pose as banks, service providers or contacts, create urgency, then request screen sharing or the installation of remote-access apps like AnyDesk or TeamViewer. Once granted, they capture OTPs, install malware or coerce transfers, enabling account takeover and financial theft. Stay skeptical: never share screens, passwords or verification codes with strangers.

🧭 This guide summarizes Multi-Agent System (MAS) fundamentals and explains how Google’s Agent Development Kit (ADK) helps developers assemble cooperating agents to solve complex tasks. It outlines three agent roles — LLM Agents for reasoning, Workflow Agents for orchestration, and Custom Agents for bespoke logic — and describes hierarchical organization and orchestration patterns (sequential, parallel, loop). The post also reviews communication options (shared state, LLM delegation, explicit invocation) and points developers to samples and codelabs for rapid prototyping.

🔒 Modern browsers include sandboxing, but attackers exploit expected behaviors to bypass protections. A new on-demand webinar from Keep Aware outlines the top three browser-layer threats—credential theft, malicious extensions, and lateral movement—and explains why tools like CASBs, SWGs, and EDRs often miss these attacks. It shows how real-time browser visibility, policy enforcement, and behavioral detection extend protection into everyday user activity. The session is aimed at CISOs and security leaders seeking practical steps to close this blind spot.

🔍 In this episode of Unlocked 403, host Becks speaks with Alena Košinárová, a software engineer at ESET, to unpack the psychological tactics behind social engineering and why people fall for scams even when they know better. They discuss how public information and social media amplify attackers' effectiveness and outline practical measures to reduce exposure. The segment balances behavioral insight with clear, actionable defenses.

🔬 Google Cloud presents how researchers can accelerate scientific workflows by combining Gemini Enterprise with integrated HPC infrastructure. It showcases AI agents—like the Deep Research agent for literature synthesis and the Idea Generation agent for proposing and ranking hypotheses—alongside developer tooling such as Gemini Code Assist and Gemini CLI for code, debugging, and workflow automation. The platform pairs these capabilities with purpose-built VMs (H4D, A4, A4X) and Google Cloud Managed Lustre to scale simulations and analysis.

🔍 This November VirusTotal is offering uncapped GUI searches for all Enterprise customers, allowing manual queries through the web interface without consuming quota. Take this opportunity to experiment with VirusTotal Intelligence search modifiers to pivot across hashes, domains, IPs, and URLs, hunt for related samples, and uncover campaign infrastructure. API interactions will continue to consume quota, while daily shared queries and community tips — tagged #MonthOfVTSearch — will help users explore advanced search techniques.

🛡️ Rapid, methodical incident response is essential when you suspect unauthorized access. Activating a rehearsed IR plan and notifying a cross-functional incident team (including HR, PR, legal and executives) helps you quickly establish scope, preserve evidence and maintain chain of custody. Contain affected systems without destroying forensic data, protect offline backups, notify regulators, insurers and law enforcement, then proceed to eradication, recovery and hardening.

🛠️ This post introduces dynamic binary instrumentation (DBI) and provides a hands-on guide to building DBI tooling using DynamoRIO on Windows 11. It explains the difference between static and dynamic instrumentation and highlights practical uses such as malware analysis, anti-anti-analysis techniques, runtime de-obfuscation, and automated unpacking. The tutorial includes example clients, build instructions, and a GitHub repository with sample code to help researchers get started.

📡 The Resilience Risk Operations Center (ROC) reframes cyber defense by fusing technical, business and financial intelligence into a single operating environment. Rather than relying solely on a traditional SOC that reacts to alerts, the ROC prioritizes threats using actuarial and claims data to show potential financial impact and guide urgent decisions. Inspired by the US Air Force AOC, it co-locates multidisciplinary experts to anticipate attacks and accelerate response. Early use, including response to an April 2024 VPN zero-day, showed faster mitigation and reduced losses.

🔒 Google Cloud's Parameter Manager centralizes application configuration to avoid hard-coded credentials and fragile config files, supporting validated JSON and YAML payloads as well as arbitrary unformatted data. It integrates with Secret Manager using a __REF__ syntax to keep confidential values separate and uses versioned, immutable parameter versions to prevent accidental changes. The post walks through storing an API key in Secret Manager, granting the Parameter Manager IAM principal access, and calling renderParameterVersion from a Node backend. A sample React/Node weather app demonstrates runtime configuration, fallback dummy data, and advanced patterns such as regional parameters and feature rollouts.