Prompt injection reached host-level code execution in Microsoft Semantic Kernel through eval() and a stray annotation
At a Glance
- Serious damage. Customer data affected, security incident, or financial losses $10k–$100k.
Independent project · aggregated from public reports and may be unverified — see the primary source below · not affiliated with or endorsed by any company or product named.
What Happened
Microsoft disclosed two vulnerabilities that turn prompt injection into host compromise in its Semantic Kernel agent framework, which has over 27,000 GitHub stars. CVE-2026-26030 affects the Python package before 1.39.4: the default in-memory vector store filter is a Python lambda executed with eval() on unsanitized model-controlled input, so an attacker could escape the template string, traverse Python's class hierarchy, bypass the AST blocklist validator and run arbitrary commands, demonstrated by launching calc.exe from a single prompt injection. CVE-2026-25592 affects the .NET SDK before 1.71.0: DownloadFileAsync was accidentally marked with a [KernelFunction] attribute, exposing it to the model with an entirely AI-controlled, unvalidated local file path, allowing writes to locations such as the Windows Startup folder and thus sandbox escape.
Case Analysis
Verified Facts
- CVE-2026-26030 stemmed from eval() on a lambda filter built with model-controlled input
- CVE-2026-25592 came from DownloadFileAsync being accidentally annotated as a [KernelFunction]
- Fixes shipped in Semantic Kernel Python 1.39.4 and .NET 1.71.0
Not Publicly Confirmed
- How many applications shipped the vulnerable defaults
- Whether either was exploited in the wild
Operational Lessons
- Any eval() or dynamic expression built from model output is a direct code execution path
- Annotation-driven tool exposure needs an audited allowlist, because one stray attribute hands the model a filesystem primitive
Primary Source
When prompts become shells: RCE vulnerabilities in AI agent frameworks (Microsoft Security Blog)microsoft.com ↗Case Record
More Cases
SearchLeak let a single click on a malicious link pull emails, files and MFA codes out of Microsoft 365 Copilot
Varonis Threat Labs disclosed CVE-2026-42824, a critical flaw in Microsoft 365 Copilot Enterprise Search. Clicking a single crafted link chained three bugs: injection of the URL parameter into the prompt, a rendering race condition, and abuse of Content Security Policy allowlisting to exfiltrate data through Bing's infrastructure. The attack could reach emails and calendar details, indexed SharePoint and OneDrive files, one-time and MFA codes, password-reset links, meeting notes and salary data. Because Copilot Enterprise is a managed service, tenant admins could not patch it themselves; Microsoft mitigated the flaw on its backend.
Samsung banned ChatGPT after engineers leaked confidential source code into it three times in 20 days
In April 2023, within about 20 days of allowing ChatGPT, Samsung's semiconductor division had three incidents of employees pasting confidential data into ChatGPT — proprietary source code to check for bugs, code for defect-detection equipment, and a recording of an internal meeting transcribed for summarization. Because prompts can be retained by the provider, this risked exposing trade secrets. Samsung banned generative AI tools company-wide and warned that violations could lead to termination.
Cursor's command allowlist could be bypassed with shell built-ins, giving prompt injection a silent path to code execution
Pillar Security disclosed CVE-2026-22708 in Cursor. In Auto-Run Mode with an allowlist enabled, shell built-ins such as export, typeset, declare, readonly, unset and local were implicitly trusted by Cursor's server-side evaluator and executed without appearing in the allowlist or requiring approval, because they run inside the shell session rather than as separate binaries. An attacker delivering indirect prompt injection could silently poison environment variables and then trigger malicious code through trusted developer tools, producing both zero-click and one-click remote code execution. Pillar reported it in August 2025, Cursor acknowledged it as a systemic issue in September 2025, and the fix shipped in version 2.3 in January 2026, which now requires explicit approval for any command the parser cannot classify.