Case
Barracuda reported that an attacker used a former AI vendor's still-active cloud access key after the vendor's development environment had been compromised. The investigation described ARTEX-pattern automated activity across cloud resources, development services and CI workflows. The first-party response does not prove every command was selected autonomously, that all possible XDR customer integration credentials were downloaded, or that Barracuda and other
Setting: productionExploitation: confirmed in the wildEvidence: Grade BRemediation: mitigation available
incidentWITH AI4 claims
Case
A deterministic agent-write policy achieved zero attacks on a bounded benchmark at the cost of excluding many legitimate tasks that need later-computed destinations.
Setting: evaluationExploitation: not establishedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 12, 2026AGAINST AI4 claims
Case
Chainstack described an intrusion that started from an exposed development Grafana instance with default credentials. Its data-source proxy could reach an internal Kubernetes MCP server without independent authentication, and commands there ran with cluster-administrator authority. The public incident account does not prove every action was model-chosen, that every accessible secret was exfiltrated or that the ARTEX tool lineage identifies the same operator a
Setting: productionExploitation: confirmed in the wildEvidence: Grade BRemediation: mitigation available
incidentWITH AI4 claims
Case
A cooperative language model encodes short identifiers through selected output-token bias, but the extraction rate falls under some editing and instruction-tuned variants.
Setting: evaluationExploitation: not establishedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 12, 2026WITH AI4 claims
Case
A small decision model assists deterministic vulnerability mining and agent review, but its poor patched-pair discrimination limits standalone use as a security verdict.
Setting: evaluationExploitation: not establishedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 12, 2026WITH AI4 claims
Case
MetaPI converts realistic agent utility tasks into adaptive prompt-injection evaluations, reusing AppWorld, VitaBench and Terminal-Bench environments. Its tests exposed cases where a defense appeared to block attack objectives while legitimate task completion also collapsed. This is a benchmark with controlled attacks and varying models. It does not establish a universal ranking of every production agent defense or that every requested harmful call was
Setting: evaluationExploitation: demonstratedEvidence: Grade BRemediation: not applicable
researchAGAINST AI3 claims
Case
An MLflow tracking API used an untrusted model identifier as a filesystem path component in FileStore, allowing out-of-run metadata writes if the service was reachable.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen Oct 26, 2026AGAINST AI4 claims
Case
MobileProbe shows that placing hostile UI content where a mobile agent will encounter it changes measured attack success, even before the payload is optimized.
Setting: simulationExploitation: demonstratedEvidence: Grade BRemediation: none known
researchFirst seen Nov 12, 2026AGAINST AI4 claims
Case
A local listwise coding agent inspects retrieved commit candidates to link CVEs to likely fixing changes, but selection accuracy does not establish that a patch is safe or complete.
Setting: evaluationExploitation: not establishedEvidence: Grade BRemediation: not applicable
researchFirst seen Nov 1, 2026WITH AI4 claims
Case
A Postgres MCP server placed arguments inside fixed dollar-quoted procedural SQL, so a crafted tool argument could become executable SQL under the connecting database role.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen Nov 11, 2026AGAINST AI4 claims
Case
A health-check error path in a Go model proxy returned upstream provider responses containing masked secret fragments to permitted health callers.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen Sep 3, 2026AGAINST AI4 claims
Case
The pgEdge Postgres MCP Server count_rows tool accepted a where expression containing a subquery against an unrelated table. Repeated counts could reveal facts about that table even when the MCP tool did not expose it by name. The upstream reproducer does not establish a production data leak or a bypass of native PostgreSQL SELECT restrictions.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
vulnerabilityAGAINST AI3 claims
Case
A tool advertised as read-only trusted statement-prefix checks rather than a transaction and role boundary, allowing writes when PostgreSQL credentials retained write grants.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade BRemediation: patch available
vulnerabilityFirst seen Sep 3, 2026AGAINST AI4 claims
Case
A colocated observer classifies a target agent's concealed approval rules from GPU timing while receiving neither its prompt nor its model responses.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade CRemediation: none known
researchFirst seen Nov 12, 2026AGAINST AI4 claims
Case
A hybrid language model can leave private attributes in recurrent state after conversation turns, while carefully controlled GPU power measurements reveal a weaker related signal.
Setting: research demonstrationExploitation: demonstratedEvidence: Grade CRemediation: none known
researchFirst seen Nov 11, 2026AGAINST AI4 claims
Case
SEAL perturbs a small selected portion of a face photo to interfere with diffusion-model personalization, but routine JPEG recompression substantially weakens the result.
Setting: evaluationExploitation: not establishedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 10, 2026AGAINST AI4 claims
Case
Researchers combine signed symbolic facts, local rule inference and private retrieval, separating authenticity, factual correctness and query privacy.
Setting: simulationExploitation: not establishedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 10, 2026AGAINST AI4 claims
Case
Keyed sampling and error-correcting text coding can transmit a small covert payload through generated language despite some model and hardware mismatches.
Setting: evaluationExploitation: demonstratedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 12, 2026WITH AI4 claims
Case
Microsoft researchers test task-purpose profiles as a defense against agent goal drift and injected instructions, without demonstrating deterministic prevention of unauthorized effects.
Setting: simulationExploitation: not establishedEvidence: Grade CRemediation: not applicable
researchFirst seen Nov 12, 2026AGAINST AI4 claims
Case
An LLM guided by known static-analysis alerts made apparently ordinary JavaScript rewrites that preserved unsafe flows while removing some CodeQL warnings.
Setting: evaluationExploitation: demonstratedEvidence: Grade BRemediation: none known
researchFirst seen Nov 12, 2026WITH AI4 claims
No matches on this page.