AI Governance Institute
← News
Enforcement2026-11-17

IBM MCP Gateway Flaw Sends User Identity to Servers Configured to Block It

What happened

IBM published a security bulletin for CVE-2026-110278 on November 12, 2026, disclosing a high-severity flaw in IBM ContextForge MCP Gateway versions 1.0.0 through 1.1.0. MCP (Model Context Protocol) gateways act as a controlled bridge between AI agents and the external tools and services those agents can call. ContextForge lets administrators configure per-server identity forwarding. Controllable details include user IDs, email addresses, group memberships, and team memberships. These settings determine what each upstream server receives. The vulnerability means those per-server settings are ignored when an agent invokes a tool or reads a resource. Identity information is forwarded to servers explicitly configured not to receive it. The flaw also involves static signature payloads. An attacker who captures an identity token can replay it to impersonate the original user. IBM assigns this a CVSS base score of 7.4. The fix is an immediate upgrade to version 1.1.1. Organizations unable to upgrade should disable identity propagation entirely by setting the global configuration flag to false. They should also restrict gateway access to trusted users. The 15,000 Public MCP Servers Found Vulnerable, Abandoned, or Sovereignty-Risky and 253 MCP Server Bugs and Six Framework Bypasses Expose Agent Authorization Gap findings form relevant background. This vulnerability adds an identity-routing failure to the growing pattern of MCP-layer control gaps.

Why it matters

  • ·Organizations that designed their MCP deployments around per-server identity controls now have no assurance those controls were enforced in production. Access records or data-minimization evidence may be inaccurate if built on assumptions about identity data routing. This creates audit exposure under the GDPR and similar frameworks requiring demonstrable data minimization.
  • ·Static identity tokens captured from gateway traffic could be reused by an attacker to impersonate legitimate users. This applies across multiple upstream MCP servers, expanding the blast radius beyond the misconfigured server boundary. This directly implicates identity and access controls for any enterprise running agentic workflows through ContextForge.
  • ·The global workaround, disabling identity propagation entirely, removes a capability that many agentic workflows depend on to enforce user-level access decisions at the tool layer. Enterprises face a forced choice between a potentially disruptive configuration change and continued exposure, a governance decision that belongs at the risk ownership level, not just with engineering teams.

Stay ahead of stories like this

Get every Global AI governance development like this one, plus the rest of the week's developments. Every Thursday.

Free every Thursday. Unsubscribe anytime.

Governance controls affected

What to do now

  • ☐Confirm with your engineering or infrastructure team whether IBM ContextForge MCP Gateway versions 1.0.0 or 1.1.0 are running anywhere in your environment, including in developer tooling and test environments, not only production.
  • ☐If affected versions are present, verify that the upgrade to version 1.1.1 has been scheduled and completed; if the upgrade cannot be done immediately, confirm that identity propagation has been disabled globally and that gateway access is restricted to known trusted users as IBM's bulletin directs.
  • ☐Review your MCP gateway configuration records to identify which upstream servers were configured to receive user identity information and which were not, then assess whether any servers in the 'blocked' category may have received identity data since version 1.0.0 was deployed.
  • ☐Determine whether any data protection obligations, particularly under privacy laws requiring data minimization, require notification or documentation updates if user identity data was sent to servers that should not have received it.
  • ☐Update your MCP server and AI gateway inventory to include version tracking and a process for receiving IBM security bulletins for ContextForge, so future patches can be assessed and scheduled without manual discovery.

What to watch next

IBM's bulletin notes the vulnerability was externally reported and covers all deployments through version 1.1.0. Organizations should monitor whether further analysis identifies additional configuration bypasses in the fixed version. MCP-layer security findings now include tool poisoning, weak authorization, and identity routing failures. This pattern is drawing attention from security agencies. Compliance teams should watch for guidance from regulators or standards bodies extending agentic AI security expectations to MCP infrastructure. Teams subject to the Five Eyes Guidance on the Careful Adoption of Agentic AI Services should consider whether gateway misconfiguration risk warrants inclusion in their agentic AI security review cycles.

Related Coverage

Corporate Policy2026-11-13

United Airlines Blocks AI Agents Over Credential Risk and Federal Aviation Rules

United Airlines announced it will block AI agents and autonomous assistants from booking flights or managing reservations on customers' behalf. The airline cited credential-sharing risks and potential violations of federal aviation rules, including requirements that a human acknowledge banned substances at check-in. United is updating its booking, ticketing, and distribution policies to cover both its own digital platforms and indirect access through travel agencies and aggregators.

Research2026-11-13

15,000 Public MCP Servers Found Vulnerable, Abandoned, or Sovereignty-Risky

OX Security analyzed over 15,000 MCP servers on public registries and 3,000 open-source AI projects, presenting findings at OWASP Global AppSec USA 2026. Researchers identified remote command execution risks, abandoned servers that attackers could take over, and data sovereignty concerns. Enterprises using AI agents that connect to MCP servers may be inheriting these risks without a systematic review process in place.

Research2026-11-17

87.5% of Known MCP Vulnerabilities Are Critical or High Severity

Researchers at IEEE SPW 2026 published a curated dataset of 24 validated MCP vulnerabilities drawn from CVE disclosures through February 2026. They found that 87.5% are Critical or High severity, and that two attack types — injection and path traversal — account for 66.7% of cases. The paper also releases a maintained dataset and reproducibility tooling for ongoing MCP security measurement.