Signed MCP tools can leak data despite signatures and runtime defences
Germline execution in ephemeral copies protects AI agent toolchains from malicious code in signed MCP tools.
Stories per day · 30 days
1 RSS
Relatedshadow-ai4agents2ai-agents2llm-agents2prompt-injection2access-control1agent-security1behavioral-analytics1benchmark-leakage1chatgpt1clinical-ai1confidentiality1dlp1exeon1germline-execution1leakbench1malware-as-image1malware-classification1
20 stories, newest first · one card per story
Stories tagged #data-leakage
Malware-as-image benchmarks leak near-duplicate samples into tests
Content-aware deduplication cuts top-1 accuracy by up to 8 points and macro-F1 by up to 15 on leaky benchmarks.
Exeon launches AgentGuard for shadow AI and autonomous agents
The product uses existing data sources, without extra agents, to flag data leakage and unexpected agent actions.
LLM agent security depends more on data access than model guardrails
Concentric AI’s chief scientist says access limits, identities, and approval gates bound what agents can leak or do.
Security teams rank employee AI data use above AI-powered attacks
In O’Reilly’s September survey, 55% had seen AI-generated phishing, but only about a quarter called it a top concern.
Agent Mosaic measures confidentiality leaks from multi-document inference
GPT-5.6-Sol disclosed forbidden composites in 7.08% of simulated answers to benign single-document queries.
LeakBench audits clinical agents for prediction-time leakage
DeepSeek Flash and GPT-6-astra leaked in 10 of 10 unconstrained September runs and in none when an admission-time constraint was set.
Employees using personal ChatGPT at work? Here's what IT admins need to do
NordLayer pitches AI Scout to block employees from pasting sensitive data into personal ChatGPT accounts.
The Hidden AI Security Risks Threatening Businesses
Executives outline business AI risks including prompt leaks, overpowered agents, shadow tools, and hiring fraud.
2
Lineage-Aware Memory Governance: A Derivation-Gated Framework for Privacy-Preserving Column-Level Access Control in Enterprise AI Agents
A lineage-gated memory policy blocks enterprise AI agents from retrieving results derived from unauthorized columns.
"Nothing to See Here'': Unintended Disclosure through Revision Traces of LLM Deliverables
LLM revision comments can leak withdrawn secrets, recoverable from about 13% of tested deliverables.
AI tests the limits of enterprise security governance
AWS’s Reimagine 2026 report says shadow AI and weak governance are outpacing enterprise security controls.
2
AgentTell: Behavioural Side-Channel Leakage in Browser-Use Agents
AgentTell shows browser-use agents leak private facts through actions in 61.1% of sessions.
PrivDrift: Auditing User-Secret Leakage Under Topic Drift in Active LLM Conversations
PrivDrift finds user secrets remain recoverable in LLM chats after topic drift, with 38.7–54.6% leakage.
Multi-View Fusion for Encrypted C2 Detection: A Leakage-Controlled Measurement Study of Evaluation Pitfalls
A leakage-controlled study of Cobalt Strike TLS flows finds fusion gains are tiny and evaluation pitfalls dwarf real detection effects.
Abandoned IoT apps keep sending sensitive data to broken servers
UMass Amherst researchers found 61,500 abandoned Android IoT apps send sensitive data to unreachable, blocklisted, or re-registered domains.
2
10 most critical LLM vulnerabilities
OWASP updated its Top 10 LLM application vulnerabilities, ranking prompt injection first and elevating excessive agency to third amid agentic adoption.
2
Shadow AI in Financial Services | Risk & Governance
Huntress warns financial services firms that unsanctioned 'Shadow AI' tool use creates data leakage and compliance risks faster than governance controls can keep pace.
4
The Outsized Shadow: Why 5% of AI Users Are Your Biggest Security Risk
Akamai's 2026 Enterprise AI Usage report finds the top 5% of AI power users create outsized shadow AI, data leakage, and agent security risks.
2
UK Legal Regulator Raises AI Misuse Concerns
UK's Solicitors Regulation Authority warns law firms about AI hallucination risks and client data leaks.