AI-Powered Cloud Attack Wipes Azure in Minutes

A threat group called JadePuffer — tracked by Microsoft as Storm-3168 and identified by Sysdig researchers as the first documented AI-driven ransomware operation — compromised a Microsoft Azure environment in early June and spent hours methodically dismantling it, according to a Microsoft Security Research blog post published on September 25.

The attacker compromised two service principals (automated accounts that cloud applications use to authenticate and perform tasks) belonging to the same Azure tenant. One spent roughly 15 hours performing reconnaissance, logging more than 300 successful read operations across virtual machines, subscriptions, and resource groups. The second then swept the same environment in five seconds flat.

Sixteen hours later, the destructive phase began. Over 100 attempts to delete storage accounts followed in rapid succession — most succeeded. A Key Vault, Function App, and App Service plan were wiped. Attempts to delete SQL databases failed only because the attacker used an outdated API version.

Microsoft found that the compromised account credentials had been posted in plaintext to a public GitHub issue by an employee of the targeted organisation. The issue was later edited to remove them, but the edit history remained publicly visible.

No ransom note was observed and data exfiltration was not confirmed — but the pattern is consistent with ransomware and extortion operations.

What to do: Search your organisation's public GitHub history for any committed or posted credentials. Rotate any service principal secrets immediately if you find them. Enable audit logging on your cloud accounts so destructive operations trigger alerts in real time.

Poper Blocker, a Chrome extension with over two million active users and a 4.8-star rating, is spyware, according to research by Bay Area Labs shared exclusively with Dark Reading.

Despite carrying Google's "Featured" badge and "Established Publisher" status, the extension exfiltrates (secretly sends out) detailed browser histories, page screenshots, and full chat logs from Claude, ChatGPT, and Gemini — including prompts, responses, subscription tiers, and model choice. It also assigns users cross-device IDs to link activity across multiple machines. The extension uses code obfuscation (deliberately scrambled code designed to hide what a program is actually doing) to conceal its behaviour.

Bay Area Labs reported Poper Blocker to Google. The app remained live as of the article's publication. Two other extensions from the same developer, also "Featured," had been flagged as spyware years prior.

What to do: Open Chrome, go to Extensions, and remove Poper Blocker if it is installed. Review all your installed extensions — delete anything you do not actively use or recognise.

What Citrix NetScaler ADC and Gateway are: NetScaler ADC (Application Delivery Controller) and NetScaler Gateway are Citrix networking appliances used by organisations to manage, secure, and control access to internal applications — often sitting at the edge of a corporate network as a gateway for remote workers.

What it is: CVE-2026-88771 is a remote code execution (RCE) vulnerability — a flaw that lets an attacker run their own commands on a system they do not own. It requires no login to trigger and affects all NetScaler ADC and Gateway deployments, including default configurations.

Who's at risk: Any organisation running an unpatched NetScaler appliance exposed to the internet. Citrix confirmed active exploitation at multiple customers worldwide, as did CISA.

CVSS: 9.5 (Critical — patch today).

Root cause: The flaw exists in a code path that can be reached without authentication. Think of it like a side entrance to a building that has no lock and no guard — the main door requires a keycard, but this entrance bypasses all of that. The developer failed to enforce authentication checks before reaching the vulnerable function.

Attack vector: An attacker with network access to the appliance sends a specially crafted request. No credentials are required. The request reaches the vulnerable code and allows the attacker to execute commands with the privileges of the running service. From there, moving through the internal network to reach other systems becomes straightforward.

Detection strategies: Check NetScaler logs for unexpected read operations or API calls from unfamiliar IP addresses. Look for unusual process spawning on the appliance. Review any recent configuration changes made outside normal change windows.

Recommended actions:

1. Apply Citrix's patches immediately — this is actively exploited and CISA has added it to its Known Exploited Vulnerabilities (KEV) catalog.
2. If patching is not immediately possible, take the appliance offline until it can be patched, as several administrators were advised to do.
3. Review Citrix's published indicators of compromise (IoCs — specific technical signs that a system has been attacked) before patching, per CISA's guidance.

The JadePuffer attack succeeded in part because credentials were accidentally posted to a public GitHub page. This is more common than most organisations realise — developers commit API keys, passwords, and service account details to public repositories regularly, often without knowing it.

A free tool called GitGuardian's Public Monitoring scans public GitHub repositories for exposed secrets and alerts you when it finds them. For a broader sweep, truffleHog is a free, open-source tool that scans repository history for credentials — including content that was edited out but remains in the commit log.

What to do: Run a search of your organisation's public GitHub repositories for your domain name, service account names, or any known credential prefixes. Rotate anything that appears. Then set up automated scanning so it never gets this far again.

A New Mexico jury found this week that Facebook violated state consumer protection law nearly 44 million times by misrepresenting how it handles user data — including false claims that users controlled their own information and that Facebook did not sell it to advertisers. Each violation carries up to $5,000 in civil penalties, meaning the final bill could reach billions. A judge will set the damages amount in the coming weeks.

For everyday users, the verdict is a reminder that platform privacy promises are worth reading critically. For businesses, it signals that state-level enforcement of data privacy claims is very much active, according to The Record.