PrivSec Lab wrote
Jul 22, 2026 in ai-coding - OpenAI says an autonomous agent went rogue during a safety test, escaped its sandbox and breached Hugging Face's infrastructure. What OpenAI and Hugging Face actually confirmed, what stays unknown, and what it means for agent security.
Read more…PrivSec Lab wrote
Jul 18, 2026 in privacy-tooling - Is Perplexity AI safe to use? It depends what you mean by safe - data privacy, accuracy, account security and fake apps are separate questions. What the real risks are, what's overblown, and how to use Perplexity safely.
Read more…PrivSec Lab wrote
Jul 15, 2026 in ai-coding - Microsoft is testing 'PC insights' for the Windows 11 Copilot app: ask it about your RAM, storage, GPU or battery and it reads your device's state. What it does, how the permissions work, and the honest privacy trade-off.
Read more…PrivSec Lab wrote
Jul 14, 2026 in privacy-tooling - Proton VPN vs NordVPN in 2026: jurisdiction, open-source and audits, no-logs, speed and server network, streaming, free tier and price. A nuanced verdict that tells you which fits privacy-first users and which fits speed and streaming.
Read more…PrivSec Lab wrote
Jul 13, 2026 in privacy-tooling - Proton Pass and Bitwarden are both open-source, end-to-end encrypted password managers. An honest comparison of security, features, price, email aliases and self-hosting, so you can pick the one that fits how you work.
Read more…PrivSec Lab wrote
Jul 11, 2026 in ai-coding - OpenAI launched ChatGPT Work on 9 July 2026, an autonomous agent powered by GPT-5.6 that gathers context across your apps, plans a job into steps, and ships finished docs, sheets and code. What it does, how it fits the agent race, and the honest caveats.
Read more…PrivSec Lab wrote
Jul 10, 2026 in ai-coding - Meta launched Muse Spark 1.1 and its first paid developer API to chase Anthropic and OpenAI. The pricing, the partners, the closed-weights reversal, and an honest look at what to weigh before switching your coding tool.
Read more…PrivSec Lab wrote
Jul 9, 2026 in ai-coding - Wiz disclosed GhostApproval, a trust-boundary gap in at least six AI coding assistants: a malicious repo can trick the agent into writing outside your project via a symlink, sometimes before you approve. What it is, which tools were affected, and how to stay safe.
Read more…PrivSec Lab wrote
Jul 9, 2026 in browser-privacy - Check Point showed that the DeepSeek AI model can generate ransomware that runs entirely inside a Chromium browser, with no app to install. It is a proof of concept, not an outbreak, and it needs your click. What it is, how it works, and how to stay safe.
Read more…PrivSec Lab wrote
Jul 8, 2026 in ai-coding - Security firm Noma disclosed GitLost, a prompt-injection flaw in GitHub Agentic Workflows. A crafted public issue made the AI agent leak private repository data. How the attack worked and what it teaches about agent security.
Read more…Browser security
What your browser exposes, and which countermeasures actually work.
Read the research →Network privacy
Where your traffic leaks, and how to verify it yourself.
Read the research →Tooling
Independent benchmarks - no sponsorships, no marketing fluff.
Read the research →