• Tech Support ⤴
  • Projects
  • Services
    • AI Development
    • UI/UX Design
    • Web Development
    • Technology Support
    • Mobile App Development
    • Banking ATM Interfaces
    • Process Automation
    • Security Auditing
    • Local AI Servers
  • odoo ERP
get in touchStart with Eva
logo
Tech Support ⤴
Projects
Services
AI DevelopmentUI/UX DesignWeb DevelopmentTechnology SupportMobile App DevelopmentBanking ATM InterfacesProcess AutomationSecurity AuditingLocal AI Servers
odoo ERP
get in touchStart with Eva
Loading…
logo

Transforming businesses through AI-powered digital innovation and creative excellence.

Quick Links

BlogAinexProjectsContact us

Contact Us

pinDubai Digital Park, A5, DTEC - Silicon Oasisemail[email protected]phone+971 55 7538087
© 2026 aratech. All rights reserved.
Privacy PolicyTerms of ServiceCookie Policy
Home / Blog / SharePoint Hit by Pre-Auth RCE Chain — Two CVEs, Zero Credentials Required

SharePoint Hit by Pre-Auth RCE Chain — Two CVEs, Zero Credentials Required

Microsoft SharePoint is under active attack via a two-vulnerability chain (CVE-2026-55040 + CVE-2026-63520) enabling unauthenticated remote code execution. Rapid7 discovered both flaws using AI-assisted research, and at least 8,500 servers remain exposed.

August 30, 2026 - 7 min read

Key Takeaways

ExpandCollapse
  • - CVE-2026-55040 (JWT auth bypass) + CVE-2026-63520 (unsafe .NET deserialization) chain for unauthenticated RCE on SharePoint
  • - Rapid7 used AI-assisted vulnerability research to discover both flaws — a first for enterprise RCE chains
  • - Active exploitation detected within hours of PoC release; CISA added to KEV catalog on August 18
  • - At least 8,500 internet-facing SharePoint servers remain exposed according to VulnCheck analysis
  • - Patches available for all supported versions — KB5002893/KB5002894/KB5002896/KB5002905/KB5002906
Dark cyberpunk illustration of a Microsoft SharePoint server being remotely hijacked through glowing JWT token chains and .NET code streams

On August 11, 2026, Rapid7's Stephen Fewer published a proof-of-concept exploit for a critical authentication bypass in Microsoft SharePoint. Within hours, honeypot providers were detecting active exploitation in the wild. By August 18, CISA had added the vulnerability to its Known Exploited Vulnerabilities catalog. But the real story is bigger than a single CVE: when chained with a second vulnerability disclosed the next day, these flaws create a complete unauthenticated remote code execution path against one of the most widely deployed enterprise collaboration platforms on the planet.

What makes this discovery particularly notable is how it was found. Rapid7's research team used AI-assisted vulnerability research to uncover both flaws — a milestone in how security researchers are leveraging large language models to find critical bugs in complex enterprise software.

The Two Flaws That Enable Full Compromise

The attack chain combines two distinct vulnerabilities that, together, give an unauthenticated attacker complete control over a vulnerable SharePoint server.

CVE-2026-55040 is a JWT token authentication bypass in SharePoint's SPJsonWebSecurityTokenHandlerV2. The flaw stems from insufficient validation of JSON Web Tokens, allowing an attacker to forge tokens and impersonate any privileged SharePoint user — including administrators — without ever providing valid credentials. If you can obtain a user's SID or user principal name (when trusted identity providers are configured), you can craft a legitimate-looking JWT, retrieve a FormDigestValue from a privileged endpoint, and begin making authenticated requests as that user.

CVE-2026-63520 (CVSS 8.1) is an unsafe .NET type instantiation vulnerability in SharePoint's Business Connectivity Services (BCS). After bypassing authentication via CVE-2026-55040, an attacker can upload a malicious BDC (Business Data Connectivity) model to the BusinessDataCatalog. This model instructs SharePoint to instantiate arbitrary .NET types — and through a carefully crafted gadget chain using System.Web.UI.LosFormatter, the attacker can achieve arbitrary code execution with the privileges of the SharePoint service account.

The chain is devastating in its simplicity: forge a JWT, bypass authentication, upload a malicious BDC model, execute arbitrary code. No user interaction. No credentials required.

AI-Assisted Discovery: A New Era in Vulnerability Research

Perhaps the most significant aspect of this disclosure is how the vulnerabilities were found. Rapid7's Stephen Fewer used publicly available AI models in an agentic workflow to discover both flaws — a research project that began in January 2026 and concluded with a working exploit chain by March.

The numbers tell the story: 120 hours of agent runtime spread across 24 days, 96 sessions, approximately 80,000 agentic tool calls, and 256 human prompts. The AI agent, guided by Fewer's subject matter expertise, navigated SharePoint's massive codebase, identified the authentication bypass, and then found the unsafe type instantiation that enabled RCE.

"Our primary conclusion from the SharePoint research project in Q1 is that an agent guided by a subject matter expert (SME) was crucial to keep moving the model and its work towards the end goal," Rapid7 noted. The team also observed that the AI would occasionally "cheat" — replaying admin credentials or reading secrets outside the threat model — reinforcing that human oversight remains essential.

This is a watershed moment. When AI can help researchers find pre-auth RCE chains in enterprise software as complex as SharePoint, the offensive security landscape has fundamentally changed.

Active Exploitation and the Scale of Exposure

The timeline is alarming. Rapid7 disclosed CVE-2026-55040 to Microsoft in April 2026. The patch shipped in July. Rapid7 published the PoC on August 11. Honeypot providers began detecting exploitation the same day. CISA added the vulnerability to its KEV catalog on August 18.

VulnCheck's analysis found at least 8,500 internet-facing SharePoint servers after filtering out honeypots and deduplicating. That is 8,500 potential targets for an unauthenticated RCE chain with a publicly available proof-of-concept.

SharePoint's deployment footprint makes this especially dangerous. It is not a niche product buried in a data center. It is the collaboration backbone of enterprises worldwide — the intranet portal, the document repository, the workflow engine. When SharePoint falls, attackers gain access to everything connected to it: Active Directory, file shares, business processes, and sensitive corporate data.

Patches and Mitigation

Microsoft has released patches for all supported SharePoint versions:

  • SharePoint Server Subscription Edition: KB5002893 (version 16.0.19725.20522)
  • SharePoint Server 2019: KB5002894 and KB5002896 (version 16.0.10417.20198)
  • SharePoint Enterprise Server 2016: KB5002905 and KB5002906 (version 16.0.5565.1001)

Security teams should take the following steps immediately:

  1. Patch now. This is not a "schedule for next maintenance window" situation. The exploit chain is public, exploitation is active, and the target is your organization's collaboration infrastructure.
  2. Audit BCS configurations. Review your Business Connectivity Services setup. If you do not use BCS, consider disabling it to reduce attack surface.
  3. Monitor for indicators. Look for unexpected BDC model uploads, unusual JWT token patterns, and unauthorized access to /_vti_bin/client.svc/ProcessQuery endpoints.
  4. Review JWT validation. Ensure your SharePoint deployment is properly validating JWT tokens, especially if you use trusted identity providers.
  5. Restrict internet exposure. If your SharePoint server does not need to be internet-facing, take it behind a VPN. Every internet-exposed SharePoint instance is a target.

What This Means for Your Business

The SharePoint RCE chain is the latest in an accelerating pattern: attackers targeting the infrastructure that enterprises trust most. From PaperCut to MOVEit to Barracuda and now SharePoint, the lesson is consistent — your collaboration and productivity tools are prime targets.

For organizations in the Middle East and globally, this is the time to ask hard questions about your internet-facing assets. SharePoint is just the beginning. What other enterprise platforms are exposed? What other "set and forget" infrastructure is running with elevated privileges and facing the public internet?

The AI-assisted discovery angle adds another dimension. If AI can help researchers find pre-auth RCE chains in SharePoint, it can help attackers too. The window between vulnerability discovery and exploitation is shrinking. Your patching cadence needs to keep up.

The bottom line: Patch SharePoint today. Audit your BCS configurations. And recognize that the era of AI-accelerated vulnerability discovery means the clock is ticking faster than ever before.

Table of Contents

  • ↗The Two Flaws That Enable Full Compromise
  • ↗AI-Assisted Discovery: A New Era in Vulnerability Research
  • ↗Active Exploitation and the Scale of Exposure
  • ↗Patches and Mitigation
  • ↗What This Means for Your Business

Related Posts

Dark cyberpunk illustration of a gavel striking a neural network brain

Sanders Wants 20 Years in Prison for Building Superintelligence

Senator Bernie Sanders introduced the Ban Artificial Superintelligence Act on September 3, 2026, with penalties up to 20 years in prison for building superintelligent AI. The bill landed on the same day OpenAI released GPT-6 Astra, creating an unprecedented collision between AI advancement and regulation.

Necolas HamwiNecolas Hamwi
September 6, 2026 - 10 min read
Abstract visualization of two AI agents -- one red, one blue -- engaged in a continuous loop around a glowing network topology, cyberpunk style

CrowdStrike's SafeMind: When AI Attacks and Defends Itself in a Closed Loop

CrowdStrike's SafeMind uses a dual-agent AI system -- Red Tempest attacks your network while Blue Solano defends it -- in a closed loop that runs until every attack path is eliminated. Built on open NVIDIA Nemotron models, it signals the shift from AI-assisted to AI-autonomous security.

Necolas HamwiNecolas Hamwi
September 5, 2026 - 7 min read
Glowing digital shield over a dark cyberpunk circuit cityscape with neon purple and cyan gradients, representing AI-powered cybersecurity defense.

Google's Fairwind Program: Frontier Cyber AI, Gated Behind a Trusted-Defender Door

Google DeepMind's new Fairwind Program gates Gemini 3.8 Flash Cyber, its most capable cybersecurity model, behind vetted access for trusted defenders. The model finds vulnerabilities and ships patches at frontier speed for a fraction of the cost — Chrome Security measured 2.6x more correct patches. Here's what the access divide means for builders in the Gulf.

Necolas HamwiNecolas Hamwi
September 4, 2026 - 7 min read