68 CVEs Uncovered in MCP Servers as 91.8% Lack OAuth Authentication, Exposing AI Agent Tool Layers
Security researchers have identified a critical blind spot in enterprise AI deployments: the Model Context Protocol (MCP) server layer that connects AI agents to external tools remains largely untracked and unprotected.
In a September report titled MCP security September 2026: Deadbugz + 3 server CVEs, Adversa AI detailed 68 vulnerabilities discovered across audited MCP servers. The AI Governance Institute followed with an analysis labeling the problem a systemic gap. Earlier data from the same researchers showed that 91.8% of examined MCP servers lacked OAuth authentication entirely.
Four Vulnerability Classes with Direct Impact
The report categorized the flaws into four well-known but newly dangerous classes in the MCP context:
- SQL injection allowing attackers to read backend databases through tool calls.
- Server-side request forgery (SSRF) specifically targeting cloud metadata services to steal temporary credentials.
- Prompt template injection that manipulates AI agents into executing attacker-chosen actions.
- Path traversal enabling unauthorized file access.
Each class has already led to data exfiltration or full agent compromise in production environments.
Additional research on 19 MCP servers revealed that tool outputs routinely inject unexpected instructions into the AI context. Context7 was specifically flagged for prompt injection issues, and Digital Applied confirmed similar behavior in its audit.
Approval Workflows and Logging at Risk
A September 4 disclosure highlighted a command classification flaw in an SSH MCP server implementation that allowed a high-privilege command to be mislabeled as safe, undermining read-only approval assumptions. Audit logs themselves become unreliable once an agent processes attacker-controlled content.
On September 10, the Cloud Security Alliance updated its guidelines, requiring OAuth 2.1 + PKCE and mandatory server metadata validation before any remote MCP connection. The guidance explicitly criticizes weak server discovery and blind trust in endpoints.
Immediate Actions Recommended
Security teams are advised to:
- Inventory all MCP servers and add them to asset management systems.
- Scan for the four vulnerability classes without sampling.
- Isolate MCP processes from cloud metadata endpoints.
- Update vendor questionnaires to cover authentication and input validation.
- Feed MCP-related CVEs into existing monitoring and watch for CISA KEV listings.
- Measure deployments against OWASP GenAI and MCP sandbox baselines.
The 68 vulnerabilities will not disappear without action. Organizations must now determine ownership of the MCP layer that currently sits outside every existing inventory and questionnaire.
Related articles
TenetBench Benchmark Tests LLMs for Compliance with Professional Ethical Codes
Researchers from the Institute of Information Security Issues at ISP RAS and the Institute of Cybersecurity and Information Protection at RANEPA have developed TenetBench, a benchmark designed to evaluate how well large language models adhere to specific norms in professional codes of ethics. The benchmark generates realistic work-related prompts that may conflict with rules from documents such as the Code of Professional Ethics for Advocates or journalistic standards on undisclosed advertising. Each test case pairs a conflicting request with a neutral counterpart to measure whether models refuse, warn, or still provide prohibited content. Evaluation separates task completion, norm violation detection, and presence of explicit refusals using an LLM-as-a-Judge approach validated against human annotators. The full dataset contains 3,433 prompt pairs derived from 15 professional codes, with a public subset of 2,235 pairs from 10 documents. Human validation confirmed 99.5% of generated examples as fully plausible work scenarios.
Solar to Showcase Solar AI Firewall for Business Protection Against AI Risks at SOC Forum 2026
GK Solar will present its new Solar AI Firewall solution at SOC Forum 2026 to help organizations control interactions with AI services and prevent confidential data leaks through employee prompts. The product addresses growing risks from staff sending corporate data to neural networks and AI agents gaining access to internal systems. The forum takes place on 27-28 October in Moscow's Lomonosov cluster, with the official launch of Solar AI Firewall scheduled during the event. Solar will also highlight practical risks of AI adoption, including data transmission to external services, use of public models by employees, and new attack surfaces in corporate infrastructure. The solution will be demonstrated as part of an integrated ecosystem for access, network, and data protection. In addition, the company plans to introduce JSOC ONE, a new line of Solar JSOC services tailored for mid-sized businesses that require 24/7 monitoring without building their own SOC.
Researcher Runs 100 Abliterated AI Agents Against His Own Accounts, Achieves Five Successful Breaches
A security researcher deployed roughly 100 autonomous AI agents based on abliterated open-source models to attempt unauthorized access to his online accounts over a five-hour period. The agents successfully compromised three accounts by exploiting vulnerabilities in legacy personal projects and two additional accounts through password reuse and credential stuffing. They also conducted 16 social engineering attempts, harvested and cross-verified personal data from public sources, and generated phishing pages targeting the researcher's Substack and Hacker News presence. Despite these successes, the agents failed to discover any zero-day vulnerabilities in third-party software or gain access to Tier-0 accounts such as Gmail, 1Password, or banking services. The experiment relied on abliterated versions of GLM-5.3 and DeepSeek V4 Flash models running on cloud GPUs, orchestrated via Codex CLI inside Docker containers with access to Railway infrastructure. Total compute cost reached approximately $210, equating to roughly $40 per successful account compromise. The researcher concluded that inexpensive swarms of such agents could soon enable low-skill attackers to systematically map and exploit an individual's entire online footprint.
How Censorship in Closed AI Models Reshapes Science: Hugging Face Releases Open Carbon-X DNA Model
Hugging Face CEO Clement Delangue announced the release of Carbon-X, an open 14.8 billion parameter language model designed for reading, analyzing, and generating DNA sequences. The announcement highlighted how shallow guardrails in closed commercial models blocked even proofreading a tweet discussing the topic. Commenters and researchers pointed out that these input/output filters lack context awareness, blocking legitimate scientific queries while motivated adversaries bypass them through rephrasing. Carbon-X was built on Qwen3-14B with a hybrid BPE and 6-mer tokenizer, domain mid-training on 302 billion tokens, and safety achieved through data curation rather than runtime censorship. The model achieves performance comparable to closed frontier systems at roughly 1/1800th the inference cost for large-scale genomic tasks. The case underscores growing asymmetry between closed corporate AI labs building wet labs and independent researchers restricted by superficial safety layers.