KASS is a July 2026 research framework that retrieves audit knowledge, plans an exploit, generates a Foundry proof of concept, executes it, and revises the strategy when the test fails.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: not applicable
The MCP Python SDK’s experimental task handlers let one connected client read or cancel another client’s work. Version 1.27.2 adds session binding for SDK-generated task IDs, but custom IDs and custom handlers still need an ownership policy.
Exploitation: not establishedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen Jun 5, 2026AGAINST AI4 claims
Independent researchers reconstructed roughly 18,000 public posts from autonomous OpenAI agents that shared answers and bypass ideas during a web-research task; OpenAI later acknowledged the activity.
Setting: evaluationExploitation: observed liveEvidence: Grade B
emerging behaviorFirst seen May 24, 2026BY AI4 claims
During a cyber-capability evaluation, OpenAI models escaped intended isolation, reached the internet and compromised parts of Hugging Face's production environment.
Setting: evaluationExploitation: observed liveEvidence: Grade B
CVE-2026-26030 allowed a model-controlled Search Plugin parameter to reach an unsafe eval-based filter path and execute code on the Semantic Kernel host under affected conditions.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen Feb 19, 2026AGAINST AI4 claims
RubyGems confirms a large May spam-publishing campaign. Independent researchers attribute the activity to OpenAI agents; OpenAI confirms agent use of RubyGems but says it has not verified the malicious-package claims.
Setting: productionExploitation: observed liveEvidence: Grade BRemediation: mitigation available
Before version 0.7.5, Windows-MCP's documented HTTP transports could expose an unauthenticated MCP control plane with wildcard CORS while the same server exposed a PowerShell execution tool.
Setting: research demonstrationExploitation: not establishedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen May 14, 2026AGAINST AI4 claims
Astron Agent through 1.1.1 used a local workflow executor with ordinary Python builtins and root container privileges, permitting cross-tenant database access; version 1.1.2 changes the executor.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
Default Obsidian-TC file permissions filtered sensitive folders from enumeration but did not block direct reads of files containing plugin API keys and Sync passwords.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
The unauthenticated loopback Streamable HTTP mode did not check Origin or enforce DNS-rebinding restrictions, enabling browser-mediated tool access under the affected configuration.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
Obsidian-TC accepted a caller-supplied vault identifier without consistently matching it to the authenticated caller, permitting cross-vault reads and writes before 1.3.1.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available