Weekly Cyber Threat Digest: Breaches, AI Exploits, and Critical Patches (April 27)

Overview

The week ending April 27 brought a flurry of cybersecurity incidents spanning data breaches, supply-chain compromises, AI-powered attacks, and urgent patches. Organizations across sectors—from cloud platforms and identity authorities to biobanks and password managers—reported unauthorized access. Meanwhile, researchers flagged novel AI exploitation frameworks and vulnerabilities in Microsoft and Apple products. Below is a breakdown of the key threats and fixes.

Top Attacks and Breaches

Vercel Breach via Context.ai Oath Token Theft

Event: Vercel, a frontend cloud platform, disclosed a security incident tied to a compromise at Context.ai. Attackers stole OAuth tokens from a connected application, enabling unauthorized access. The breach exposed employee information, internal logs, and a subset of environment variables. However, Vercel emphasized that the most sensitive secrets were not accessed. The incident highlights risks of third-party integration chain vulnerabilities.

France Titres Data Breach: Personal Data for Sale

Event: France Titres, the authority responsible for identity and registration documents, detected a data breach on April 15. The exposed data may include names, birth dates, email addresses, login IDs, and some physical addresses and phone numbers. A hacker has already offered purported agency data for sale on the dark web. The breach underscores the ongoing threat to government-adjacent databases.

UK Biobank Breach: 500,000 Volunteers' Data Advertised

Event: UK Biobank, a major research organization, confirmed a breach after de-identified health data on half a million volunteers was advertised for sale on Chinese marketplaces. Officials stated that the listings were removed and likely unsold, but immediate countermeasures included suspending access, shutting down the research platform, and imposing download limits. This incident raises concerns about the security of large-scale health research repositories.

Bitwarden Supply-Chain Attack via Malicious CLI Release

Event: Bitwarden, a widely used password manager, suffered a supply-chain attack after a malware-tainted CLI release was published to npm on April 22. The compromised version (2026.4.0) was downloaded by 334 developers during a brief window, potentially exposing credentials. Bitwarden attributed the incident to a hijacked GitHub account. Importantly, vault data—the core encrypted password storage—remained unaffected, limiting the scope of exposure.

AI Threats

Unauthorized Access to Anthropic's Claude Mythos Preview

Event: Researchers flagged unauthorized access to Anthropic’s Claude Mythos Preview, an unreleased AI cyber model. Access was gained through a third-party vendor environment. A small Discord group reportedly used shared contractor accounts, API keys, and predictable URLs to reach the system. Anthropic noted that core systems were unaffected and the incident is under investigation. This case illustrates the risks of early-access AI deployments and insufficient vendor segmentation.

Bissa Scanner: AI-Assisted Mass Exploitation Platform

Event: Researchers observed an AI-assisted exploitation platform called Bissa Scanner, which leverages Claude Code and OpenClaw to automate mass scanning, exploitation, and credential harvesting. The operation specifically targeted the React2Shell vulnerability (CVE-2025-55182). The platform scanned millions of targets, confirmed over 900 compromises, and collected tens of thousands of exposed environment files. This marks a worrying evolution in automated, AI-driven attack tools.

Prompt Injection Chain in Google's Antigravity IDE

Event: Researchers highlighted a prompt-injection exploit chain in Google’s Antigravity agentic IDE that enabled sandbox escape and remote code execution. The flaw abused a file search tool that executed before security checks, allowing attackers to convert a benign prompt into system compromise—even in Secure Mode. Google has since patched the vulnerability. This demonstrates how agentic AI tools can introduce novel attack surfaces.

Vulnerabilities and Patches

Microsoft: Critical ASP.NET Core Privilege Escalation (CVE-2026-40372)

Event: Microsoft issued out-of-band fixes for CVE-2026-40372, a critical privilege escalation flaw in ASP.NET Core rated 9.1 (out of 10). The bug affects Data Protection versions 10.0.0 to 10.0.6. It allows attackers to forge cookies and antiforgery tokens, impersonate users, and gain SYSTEM-level access on Linux or macOS deployments. Administrators should prioritize patching these versions immediately.

Apple: iOS/iPadOS Notification Services Bug (CVE-2026-28950)

Event: Apple released fixes for CVE-2026-28950 in iOS and iPadOS, a vulnerability in Notification Services. While specific impact details are still emerging, the severity warranted immediate patching. Apple users are advised to update their devices to the latest available versions.

Conclusion

The week’s events underline the importance of securing third-party integrations, monitoring AI-specific attack vectors, and applying patches promptly. Organizations should review their OAuth practices, enforce strict access controls for vendor environments, and update systems to mitigate the latest threats.