IAM & Security Weekly Briefing

Week of: July 12–18, 2026 Reporting window: Most recently completed Sunday–Saturday (excludes the in-progress week).

---

1. Executive Summary (TL;DR)

• Microsoft ships a record 622+ patches in one Patch Tuesday, with two zero-days under active attack and AI-assisted vulnerability discovery driving the surge.
• Identity attacks have overtaken exploits as the top root cause of ransomware, with MFA present in 97% of credential-based attacks yet failing to prevent compromise.
• CISA's postmortem on a contractor's GitHub leak of AWS GovCloud keys highlights systemic gaps in credential governance for non-human identities.
• The NadMesh botnet hunts exposed AI services, harvesting 3,811 unique AWS keys and Kubernetes tokens — a new frontier in cloud credential theft.
• A new "Agent Data Injection" attack technique demonstrates how AI agents can be manipulated into executing attacker commands by corrupting trusted data sources.
• SonicWall SMA 1000 zero-days (CVSS 10.0) are being actively exploited by Inc Ransomware, with root-level access achieved via chained vulnerabilities.
• WordPress Core "wp2shell" RCE flaws now have public exploits, making unpatched 6.9 and 7.0 sites trivially exploitable by unauthenticated attackers.
• EU orders Google to open Android's mic, camera, and screen to rival AI assistants, reshaping the identity and access landscape for mobile AI agents.

---

2. Top IAM & Security News

Microsoft Patches Record 622 Flaws, Including Two Zero-Days Under Active Attack

• Source: The Hacker News
• Link: The Hacker News
• Date: July 14, 2026
• What happened: Microsoft released its largest Patch Tuesday ever, addressing 622 CVEs including two zero-days already being exploited in the wild, with AI-assisted discovery driving the record volume.
• Why it matters: The sheer volume demands automated triage; IAM teams must prioritize the two actively exploited flaws and ensure identity infrastructure (Entra, AD, ADFS) is patched first.

Identity Attacks Overtake Exploits as Top Ransomware Cause

• Source: Dark Reading
• Link: Dark Reading
• Date: July 15, 2026
• What happened: Email-based identity attacks have surpassed technical exploits as the primary ransomware vector, with MFA deployed in 97% of credential-based attacks yet failing to stop compromise.
• Why it matters: MFA alone is insufficient — organizations must deploy phishing-resistant MFA (FIDO2/passkeys), conditional access policies, and continuous session validation.

Lessons Learned from CISA's Recent GitHub Leak

• Source: KrebsOnSecurity
• Link: KrebsOnSecurity
• Date: July 13, 2026
• What happened: A CISA contractor published dozens of internal credentials — including AWS GovCloud keys — in a public GitHub repository for nearly six months before being notified.
• Why it matters: Non-human identity (NHI) governance is broken across government and industry; enforce secret scanning, short-lived credentials, and automated revocation for all machine identities.

New NadMesh Botnet Hunts Exposed AI Services for Cloud Keys and Kubernetes Tokens

• Source: The Hacker News
• Link: The Hacker News
• Date: July 17, 2026
• What happened: A Go-based botnet scans for exposed AI services (ComfyUI, Ollama, n8n) and has already harvested 3,811 unique AWS keys and Kubernetes tokens.
• Why it matters: AI/ML infrastructure is a new crown jewel for credential theft; enforce network segmentation, require workload identity federation, and never embed cloud keys in AI service configurations.

WordPress Core "wp2shell" RCE Flaws Get Public Exploits, Patch Now

• Source: BleepingComputer
• Link: BleepingComputer
• Date: July 18, 2026
• What happened: Public proof-of-concept exploits have been released for critical unauthenticated RCE vulnerabilities in WordPress Core affecting versions 6.9 and 7.0.
• Why it matters: Any WordPress site — even with zero plugins — is exploitable; IAM teams managing WordPress-based portals or customer-facing apps must patch immediately or risk full site takeover.

Two SonicWall SMA 1000 Zero-Days Exploited, One Could Enable Admin Commands

• Source: The Hacker News
• Link: The Hacker News
• Date: July 15, 2026
• What happened: SonicWall warned of active exploitation of two SMA 1000 zero-days (CVE-2026-15409, CVSS 10.0) that when chained grant root-level access; Inc Ransomware is using them.
• Why it matters: VPN/remote access appliances are critical identity infrastructure; any organization using SMA 1000 series must isolate or patch immediately as these are gateways to the corporate network.

CISA Adds Exploited SharePoint RCE Zero-Day CVE-2026-58644 to KEV

• Source: The Hacker News
• Link: The Hacker News
• Date: July 17, 2026
• What happened: CISA added a critical SharePoint Server deserialization RCE (CVSS 9.8) to its Known Exploited Vulnerabilities catalog, requiring federal agencies to patch by July 19.
• Why it matters: SharePoint is often tied to Entra ID/AD for authentication; a successful RCE can lead to credential theft and lateral movement across the identity estate.

---

3. AI, Identity & Emerging Tech

New Agent Data Injection Attack Can Make AI Agents Misclick or Run Attacker Commands

• Source: The Hacker News
• Link: The Hacker News
• Date: July 16, 2026
• What happened: Researchers demonstrated that planting malicious data in sources AI agents trust (product reviews, GitHub threads) can make agents execute unintended actions — including running attacker commands.
• Why it matters: As organizations deploy AI agents with identity and access to enterprise systems, data injection becomes a new identity-based attack vector that bypasses traditional access controls.

Researchers Say Claude for Chrome Flaw Lets Rogue Extensions Trigger Gmail Reads

• Source: The Hacker News
• Link: The Hacker News
• Date: July 14, 2026
• What happened: A flaw in Claude for Chrome allows other browser extensions to trigger AI agent tasks that read Gmail, Google Docs, and Calendar — even after Anthropic's earlier fix for the PromptFiction vulnerability.
• Why it matters: AI browser extensions inherit the user's identity and access; organizations must inventory and restrict AI extensions as part of their identity security posture.

OpenAI's GPT-Red Automates Prompt Injection Testing to Harden GPT-5.6

• Source: The Hacker News
• Link: The Hacker News
• Date: July 16, 2026
• What happened: OpenAI disclosed GPT-Red, an internal automated red-teaming model that scales prompt injection vulnerability discovery and is used to adversarially train GPT-5.6.
• Why it matters: Prompt injection is the identity bypass of the AI era; organizations consuming AI APIs should demand transparency on red-teaming practices and implement their own prompt validation layers.

Google Bets 'Agentic Defense' Strategy Can Outpace Attackers

• Source: Dark Reading
• Link: Dark Reading
• Date: July 17, 2026
• What happened: Google Cloud is incorporating Wiz capabilities into an agentic defense platform that automates threat detection and remediation against AI-powered attacks.
• Why it matters: Agentic defense represents a shift from static IAM policies to autonomous identity threat detection and response — a trend IAM teams must prepare for.

1M+ Emails Use Hidden Text to Dupe AI Security Filters

• Source: Dark Reading
• Link: Dark Reading
• Snippet: AI and LLMs can be surprisingly ineffective against text salting, allowing phishing emails to slide right into your inbox.
• Date: July 16, 2026
• What happened: Over one million phishing emails used "text salting" — hidden or obfuscated text — to bypass AI-powered email security filters.
• Why it matters: AI-based phishing detection is not foolproof; IAM teams must layer identity-aware defenses (phishing-resistant MFA, anomaly detection) on top of email security.

---

4. Cyber Threats & Attack Trends

ACR Stealer Uses ClickFix Lures to Steal Browser Tokens and Microsoft 365 Files

• Source: The Hacker News
• Link: The Hacker News
• Date: July 17, 2026
• What happened: Microsoft warns of a surge in ACR Stealer attacks that steal browser-stored passwords, session tokens, and Microsoft 365 documents (OneDrive, SharePoint) via ClickFix social engineering.
• Why it matters: Session token theft bypasses MFA entirely; enforce token binding, short session lifetimes, and device compliance checks for all cloud app access.

New ClickLock macOS Stealer Kills Apps Every 210ms Until Victims Type Their Password

• Source: The Hacker News
• Link: The Hacker News
• Date: July 16, 2026
• What happened: A new macOS infostealer kills critical apps (Finder, Dock, Terminal) every 210ms until the victim types their login password into a fake system dialog.
• Why it matters: macOS is increasingly targeted; enforce application allowlisting, restrict Terminal/script execution for standard users, and deploy EDR on all Mac endpoints.

Two Scattered Spider Hackers Get 5.5 Years Each for £29 Million TfL Hack

• Source: The Hacker News
• Link: The Hacker News
• Date: July 16, 2026
• What happened: Two teenagers were sentenced to 5.5 years each for the 2024 Transport for London hack that left 148 systems inoperable and forced all 27,000 employees to reset passwords in person.
• Why it matters: The TfL attack relied on credential theft and social engineering; the manual password reset requirement underscores the operational chaos when identity infrastructure is compromised.

Inc Ransomware Exploits SonicWall SMA Zero-Days

• Source: Dark Reading
• Link: Dark Reading
• Date: July 17, 2026
• What happened: The Inc ransomware group is actively exploiting the SonicWall SMA 1000 zero-days (CVSS 10.0) to gain root-level access to victim networks.
• Why it matters: VPN appliances are identity gateways; a root-level compromise means attackers can manipulate authentication, intercept credentials, and move laterally undetected.

Fake Coding Tests Deliver OtterCookie-Aligned Malware Hidden in SVG Flag Images

• Source: The Hacker News
• Link: The Hacker News
• Date: July 17, 2026
• What happened: North Korean threat actors used fake job postings and coding challenges to deliver OtterCookie malware via steganography in SVG images, targeting browser credentials and crypto wallets.
• Why it matters: Developer-targeted social engineering is rising; enforce strict network segmentation for development environments and monitor for anomalous outbound connections from dev machines.

---

5. Product Updates & Vendor News

Cursor Flaw Lets Malicious Cloned Repositories Trigger Windows Code Execution

• Source: The Hacker News
• Link: [The Hacker News](https://thehackernews.com/2026/07/cursor-flaw-le