AI
The Engineer's Guide to the EU AI Act
A practical guide for engineers on translating the EU AI Act's legal requirements into technical implementation and compliance strategies.
The European Union's Artificial Intelligence Act is the world's first comprehensive legal framework for AI, now in full effect across the EU. It imposes binding obligations on providers and deployers of AI systems, categorized by a risk-based approach: unacceptable (banned), high, limited, and minimal. For engineering teams, compliance is a core operational reality, directly shaping system architecture, data governance, and MLOps practices.
This guide is designed for the engineers, MLOps professionals, and technical leaders actively building and maintaining compliant AI systems. We deconstruct the Act's key requirements into actionable technical guidance, covering data governance, model transparency, robust risk management, and the technical prerequisites for conformity assessments and post-market monitoring. The focus remains on practical implementation, helping you navigate the established standards and practices for building AI that is safe, transparent, and lawful.
Latest briefings on The Engineer's Guide to the EU AI Act
AI
Security Concerns Now Slow AI Adoption
A new Linux Foundation report finds that security readiness is the biggest obstacle to AI adoption. A widening gap exists between the rush to deploy AI and the ability to secure it. The report notes 67% of teams face pressure to accelerate deployment despite security risks.
Neeraj Dhiman ·
Security
Four Malicious npm Packages Discovered
Cybersecurity researchers have identified four malicious packages on the npm registry: `chalk-tempalte`, `@deadcode09284814/axios-util`, `axois-utils`, and `color-style-utils`. These packages were designed to steal information from developer systems and have been downloaded thousands of times.
Neeraj Dhiman ·
Infra
AWS Is Unplugging Its European Cloud for Hours
AWS will test its European Sovereign Cloud by disconnecting it from the global network for several hours. The drill aims to prove the platform's independence, a key requirement for customers navigating strict EU data laws.
Ashish Kale ·
Data
PostgreSQL Email Tool Gets a Major Security Upgrade
A popular PostgreSQL extension for sending emails directly from the database now supports encrypted connections and asynchronous sending. The update makes it a more secure and performant option for developers, especially those migrating applications from Oracle.
Taranpreet Singh ·
AI
Why Physical AI Is Still in Its Lego Era
A former Nvidia robotics lead argues that Physical AI is still in its 'Lego era.' The basic hardware and software building blocks for robotics are not yet mature, challenging the idea that a single large AI model can solve everything.
Neeraj Dhiman ·
Infra
A New Tool Aims to Unify JavaScript Development
The new JavaScript toolchain Vite+ has reached its stable 1.0 release. Backed by Cloudflare, it aims to simplify complex development workflows by managing Node.js, packages, and frontend tasks from a single configuration file.
Ashish Kale ·
Infra
A New Guide to Building Software That Evolves
InfoQ has released a free guide on Evolutionary Architecture, a method for designing software that can easily adapt to new technologies and business needs. It offers practical advice for building systems that last.
Ashish Kale ·
AI
AWS Releases Tools for Safer AI Agent Development
AWS has released two open-source tools for AI agent development. Strands Decider 2B is a small model for making decisions, while Strands Box provides a secure sandbox for running agent-generated code, addressing key safety and efficiency challenges.
Neeraj Dhiman ·
AI
AI Security Is Really Just Kubernetes Security Now
The security of AI systems now depends heavily on the underlying Kubernetes infrastructure. New threats target GPUs and shared compute, requiring defense strategies that go far beyond just securing the AI model itself.
Neeraj Dhiman ·
AI
AI Use Puts DraftKings' State License on Hold
Louisiana regulators gave DraftKings a temporary license extension to investigate its AI use. The move signals growing scrutiny for companies deploying AI in regulated industries, highlighting new operational and compliance risks for tech leaders.
Neeraj Dhiman ·
AI
How to Find Flaws AI-Generated Code Misses
AI-generated code can introduce subtle but critical system flaws. A new approach uses formal methods, a rigorous way to mathematically prove system correctness, to verify AI code before it reaches production.
Neeraj Dhiman ·
Tech
Rocket Lab Delays Its Carbon Fiber Rocket Again
Rocket Lab's medium-lift Neutron rocket is now delayed until 2027. The company cited manufacturing issues with the rocket's advanced carbon composite structures, underscoring the challenge of pioneering new materials in aerospace engineering.
Taranpreet Singh ·
Infra
Vercel Lets AI Agents React to Any App
Vercel has launched public beta support for MCP Events, a standard for interoperable communication. This lets developers build AI agents and apps that can automatically react to events, like new messages or issues, from various collaboration tools.
Ashish Kale ·
Tech
Amazon Challenges AI Hardware Market at $75
Amazon's Bee unit has launched two personal AI devices starting at $75. The new wearable and recorder summarize conversations and connect to Alexa, aiming to make AI a low-cost, everyday utility for a mass audience.
Navdeep Kaur Mahal ·
AI
AI Can Find Your Home From an Indoor Photo
A new AI tool called Raven can identify a specific home using only visual clues from an indoor photo, without any location data. This creates a powerful new tool for surveillance and a major privacy risk for everyone.
Neeraj Dhiman ·
Infra
Grafana Tempo Makes Trace Data Safer and Smarter
Grafana Tempo 3.1 now lets teams remove sensitive data from system traces using simple queries. The update also makes performance metrics more accurate by accounting for sampled data, improving both privacy and observability.
Ashish Kale ·
Data
AI Agents Are Driving a New Database Approach
A new database engine called CedarDB supports both fast transactions and complex analytics in one system. Its hybrid approach is gaining traction as AI agents need to access both operational and historical data at the same time.
Taranpreet Singh ·
Data
New Tool Brings Version Control to PostgreSQL Schemas
A new open-source tool, Perseid, brings Git-like version control to PostgreSQL database schemas. It helps development teams track, manage, and deploy database changes more reliably, reducing errors and improving collaboration on complex projects.
Taranpreet Singh ·
Data
New Tool Treats Your Postgres Schema Like Code
A new open-source tool, pglifecycle, lets developers manage PostgreSQL schemas declaratively using version-controlled YAML files. This approach aims to replace traditional, complex migration scripts, making database changes easier to review and deploy in modern CI/CD pipelines.
Taranpreet Singh ·
Data
One Password Could Compromise Your Entire Postgres DB
A critical security flaw in PgBouncer, a popular PostgreSQL tool, allows an attacker with one user's password to log in as any other user, including administrators. The new 1.26.1 release fixes this and two other vulnerabilities.
Taranpreet Singh ·
AI
Why Bad Data Is Stalling the UK's AI Boom
UK businesses now spend an average of £20.6 million on AI annually, a 27% jump. However, two-thirds of firms say poor data quality is the biggest obstacle preventing them from seeing a full return on their investment.
Neeraj Dhiman ·
Tech
Neuralink Rival Implants First Human Vision Chip
Science Corp, a startup from Neuralink's ex-president, has implanted its first vision-restoring chip in a human patient. The device aims to treat degenerative eye diseases and is a step toward a broader brain-computer interface.
Taranpreet Singh ·
AI
Cerebras CEO: AI Doomsday Talk Hurts the Industry
Cerebras CEO Andrew Feldman is urging AI leaders to stop focusing on doomsday scenarios. He argues the constant risk-talk overshadows AI's huge potential benefits and misleads the public conversation about the technology's future.
Neeraj Dhiman ·
AI
AI Job Cuts Are Moving Into the Corporate Office
Kroger, a major U.S. grocer, is cutting 3,000 corporate positions over three years. The company is using AI and other technology to streamline operations, signaling a major shift in how large, non-tech companies are restructuring their workforce.
Neeraj Dhiman ·
Infra
Backstage Update Recovers Failed Template Jobs
Backstage version 1.55 introduces a new feature that can recover failed software template jobs, improving reliability. However, the update also includes breaking changes, requiring teams to review their setups before upgrading.
Ashish Kale ·
Infra
Vercel Just Slashed Your Build Wait Times
Vercel has optimized its build process, making repository checkouts and Git LFS downloads significantly faster. The platform reports an average 20% reduction in the time it takes to prepare project files, accelerating CI/CD pipelines for developers.
Ashish Kale ·
Data
PostgreSQL Vector Search Just Got Much Faster
A new open-source extension for PostgreSQL called pg_vectorsearch dramatically improves vector search performance. It uses a novel index to deliver faster queries with significantly less memory, making AI applications more efficient and affordable.
Taranpreet Singh ·
Infra
How Workday Builds Its AI Interface On The Fly
Workday is dynamically building the user interface for its new AI assistant, Sana. The company uses a 'decision model' to generate the UI on the fly, a novel approach for enterprise software development.
Ashish Kale ·
Infra
GitLab's New AI Fixes Your Slowest Bottleneck
GitLab is using AI agents to automate the slow, manual steps after code is written, like testing and reviews. This new agent orchestration aims to eliminate the human wait times that currently bottleneck software delivery.
Ashish Kale ·
Infra
Cloudflare's New Privacy Tool Blocks Its Own Traffic
Cloudflare launched a new privacy gateway using Oblivious HTTP. Its key feature is a built-in rule that prevents it from decrypting its own traffic, architecturally guaranteeing that user identity and request data remain separate.
Ashish Kale ·
Frequently asked questions
What is the EU AI Act's risk-based approach in practice?
The Act categorizes AI systems into four risk tiers with obligations scaling accordingly. High-risk systems, like those in medical devices or critical infrastructure, must undergo conformity assessments and meet strict requirements for data quality, documentation, and post-market monitoring. Limited-risk systems, such as chatbots or AI-generated content tools, face transparency obligations, while minimal-risk systems are largely unaffected.
As an engineer, what are my key responsibilities for a 'high-risk' AI system?
Your core responsibilities include implementing a risk management system compliant with Article 9, ensuring training data is relevant and free of bias, and maintaining detailed technical documentation as per Annex IV. You must also build systems with automatic logging for traceability and design for high levels of accuracy and cybersecurity, often by adhering to the latest CEN-CENELEC harmonised standards.
How does the AI Act impact the use of open-source AI models?
Providers of general-purpose AI (GPAI) models, including open-source ones, must provide technical documentation and usage instructions to downstream integrators. While a narrow exemption exists for open-source models not classified as having 'systemic risk,' the provider integrating any GPAI model into a high-risk system always bears full responsibility for the final system's compliance.
What technical documentation is required for compliance?
For high-risk systems, you must maintain living technical documentation (per Annex IV) that details the system's architecture, intended purpose, and limitations. It must also describe the design specifications, data requirements, and the validation and testing procedures used to ensure performance. This documentation is a critical prerequisite for the conformity assessment and is subject to review by market surveillance authorities.