Weekly Recap: Vercel Hack, Push Fraud & Android RATs

A packed week in security. The Vercel hack, a surge in push notification fraud, QEMU being weaponized as a network tunneling tool, and fresh Android RAT families hitting the wild. These are not isolated incidents. They are signals of a threat environment where attackers are moving faster and using more creative infrastructure abuse than most security teams are tracking.

Vercel Infrastructure Targeted in Active Attack Campaign

Attackers went after Vercel deployments this week. The specifics involve manipulating environment variable exposure and misconfigured project settings to extract secrets from CI/CD pipelines. If you are shipping code through Vercel and your build environment has access to API keys, database URLs, or OAuth credentials, those are the targets.

The attack chain is straightforward. Gain access to a project with loose collaborator permissions, inject or read environment variables, then pivot using those credentials elsewhere. The blast radius depends entirely on how tightly scoped your secrets are. Many developers give their build environments far more privilege than the deployment actually needs.

Check your Vercel project settings. Rotate any secrets that have been sitting in environment variables for more than 90 days without audit. Scope credentials to the minimum required access.

Push Notification Fraud Scales Up with AI Assistance

Push fraud is not new. But the volume and targeting precision have jumped significantly. Attackers are now using AI to craft push notification lure content that closely matches the app's genuine notification style, making users far more likely to approve MFA prompts they should reject.

This directly connects to findings in the Zscaler ThreatLabz 2026 VPN Risk Report. AI has collapsed the human response window. Where a suspicious push might have given a user pause two years ago, the prompts now look and feel legitimate enough that approval rates are climbing. Remote access systems are the primary target because a single approved push can hand an attacker full network presence.

QEMU Abused as Covert Network Tunnel

Security researchers confirmed active campaigns using QEMU, the open-source machine emulator, to create hidden network tunnels inside compromised environments. Attackers run a lightweight virtual machine inside the victim's infrastructure, then use QEMU's networking stack to route traffic out in a way that bypasses many traditional endpoint detection tools.

This is notable because QEMU is a legitimate sysadmin tool. Detection based purely on process whitelisting will miss it. You need behavioral network analysis looking at unusual outbound connections originating from virtualization processes.

New Android RATs: What Developers Should Know

Two new Android remote access trojan families were spotted this week. Both use overlay attacks to steal credentials and both are distributed through phishing sites mimicking legitimate app stores. For developers building Android apps, this raises a direct concern. If your app handles authentication or payment flows, overlay attacks can silently intercept user input before it ever reaches your code.

Mitigations include implementing FLAG_SECURE on sensitive UI windows to block screen capture and overlay rendering, and validating that your authentication flows detect when another app is drawing on top of yours.

How to Protect Your Applications Right Now

Actionable steps from this week's events:

• Audit Vercel environment variables and rotate stale credentials immediately.
• Enable phishing-resistant MFA (hardware keys or passkeys) wherever push-based MFA is currently used.
• Add QEMU and other virtualization binaries to your behavioral monitoring watchlists.
• Review your Android app's overlay attack surface and apply FLAG_SECURE to all sensitive screens.
• Run an external scan of your web properties to catch exposed secrets or misconfigured endpoints before attackers do. Start at /scan.

Read more about securing CI/CD pipelines and secrets management on the VibeWShield blog.

FAQ

How do I know if my Vercel project was affected by this week's attack? Review your audit logs in the Vercel dashboard for unexpected collaborator additions or environment variable reads. Rotate all secrets as a precaution if anything looks off.

Is QEMU dangerous to have on developer machines? QEMU itself is legitimate software. The risk is in compromised environments where attackers already have a foothold and use it to establish persistent, hard-to-detect tunnels. Keep it off production systems if you are not actively using it.

What is the fastest way to test if my app leaks secrets externally? Run an automated DAST scan against your deployed endpoints. Many secret exposures appear in HTTP responses, error messages, or misconfigured headers that only show up under active testing.

Scan your web application for exposed secrets and misconfigurations now at VibeWShield /scan.