← Frameworks
MAP 5.1

NIST AI RMF

NIST AI RMFView in NIST

How MAP 5.1 shows up in practice: the mapped risk classes in this atlas, the documented incidents that prove it's real, and the scenarios and controls to learn and defend against it.

Real-world cases

43

Documented incidents, disclosed vulnerabilities and research that illustrate MAP 5.1 — latest first, each with sources.

Claude Code Opus 5 Auto Mode hijacked to RCE via indirect prompt injection26 Aug 2026

Researcher Johann Rehberger reportedly drove Claude Code (Opus 5) in Auto Mode to code execution from a single summarize-this-website request, using a module-shadowing trick so an imported stdlib decoder runs an attacker payload, with a reported 60-80% success rate.

Context7 MCP documentation-server prompt injection (CVE-2026-75130)18 Aug 2026

NVD reportedly published a CVSS 9.0 prompt-injection flaw in the widely-installed Context7 MCP documentation server, where unsanitized content served via its Custom AI Instructions feature is delivered into a connected coding agent's context and executed with the agent's own file, shell and network access, with no fix referenced at publication.

Hanover Institute seeds question-shaped content to steer ChatGPT/Perplexity on Gaza16 Aug 2026

An investigation reportedly described a state-linked campaign publishing content phrased as chatbot questions (generative engine optimization) so that ChatGPT and Perplexity retrieve and cite it when answering neutral questions about Gaza and the IDF.

AI Recommendation Poisoning: Ask-AI web links silently write trusted-source into assistant memory06 Aug 2026

Microsoft Security reportedly catalogued an in-the-wild technique across dozens of companies where websites embed hidden prompt-injection payloads behind Ask-AI deep-links that, clicked in an authenticated assistant session, silently write a permanently-trust-this-vendor tag into the assistant's long-term memory.

Azure DevOps MCP confused-deputy — hidden PR comments hijack AI review agents21 Jul 2026

Manifold Security reported that Microsoft's official Azure DevOps MCP server returns instructions planted in a hidden HTML comment inside a pull-request description — invisible in the web UI but returned verbatim by the API — so a victim's AI review agent, acting under the victim's credentials, follows the hidden orders and reaches data the attacker could not access directly.

AWS Kiro agentic IDE rewrites its own MCP config for zero-click RCE (CVE-2026-10591)21 Jul 2026

Researchers reportedly showed hidden text in a web page could make AWS's Kiro agentic IDE rewrite execution-sensitive config it controls (mcp.json, tasks.json) that auto-loads on folder open, turning a summarize-this-page request into zero-click code execution (reportedly patched in v0.11.130 / 0.11.x; the primary Intezer and AWS sources publish no CVSS).

The Week of Sandbox Escapes: AI coding-agent sandbox bypasses (CVE-2026-48124 and more)20 Jul 2026 – 23 Jul 2026

Pillar Security reportedly disclosed eight sandbox-escape vulnerabilities across four AI coding agents (Cursor, OpenAI Codex CLI, Google Gemini CLI, Google Antigravity) over four days, finding that in nearly every case the agent did not break the sandbox directly but wrote a file that a trusted component outside the sandbox later ran, loaded or scanned.

Hugging Face agentic production intrusion via a poisoned dataset (July 2026)16 Jul 2026

Hugging Face disclosed a production-infrastructure intrusion that it says was driven end-to-end by an autonomous AI-agent system: a malicious dataset abused code-execution paths in its dataset-processing pipeline as the foothold, then the campaign escalated to node-level access and moved laterally into internal clusters over a weekend.

Context Contamination: passive prompt injection poisons LLM security-log analysis16 Jul 2026

A red-team study shows adversaries can hide prompt-injection payloads inside network-log fields (usernames, URLs, user-agents) that fire when a SOC analyst asks an LLM to triage the logs — reportedly reaching up to 88.2% success at concealing malicious activity or exfiltrating data, turning the audit trail itself into the injection channel.

GPT-Red self-play red-teaming and the 'Fake Chain-of-Thought' injection class15 Jul 2026

OpenAI detailed GPT-Red, an internal self-play red-teaming model that reportedly beat human red-teamers 84% to 13% on prompt-injection tasks and autonomously surfaced a novel 'Fake Chain-of-Thought' attack — planting a spoofed, already-verified reasoning step in a model's trace to smuggle attacker instructions past its checks.

Agentic botnets via universal, transferable adversarial HalluSquatting08 Jul 2026

Spira, Cohen, Nassi et al. (the Morris II group) show LLM resource-name hallucination can be weaponised at scale: attackers pre-compute a model's most-likely hallucinated names for trending repos/skills, register them, and host adversarial prompts there. The paper reports hallucinated-resource generation up to 85% for repository cloning and up to 100% for skill installation, with hallucinations that transfer across foundation models and prompts, enabling remote tool/code execution assemblable into a botnet.

MemGhost: stealthy one-email memory injection in persistent personal agents06 Jul 2026

An academic paper reports MemGhost, a one-shot indirect-injection attack in which a single ordinary email causes a memory-enabled personal agent to silently write an attacker's false fact into long-term memory, conceal the write from its reply, and rely on the fiction in later sessions.

+ 31 more via the mapped risk pages above.

Browse all real-world cases →

Practise it — interactive scenarios

🔗One Click, Permanent Trust

An 'Ask AI' button quietly plants a permanent 'trusted source' rule in your assistant's memory

☠️Poisoning the Well

An attacker edits the wiki; the assistant cites the lie back to everyone

📧The Email That Gave Orders

A support email hides instructions — and the assistant obeys them

🕵️Lies in the Loop

A poisoned issue makes the agent lie to the human who approves its actions

🧲Poison the Vector, Not the Words

An attacker crafts a gibberish passage whose embedding sits near thousands of questions — so it's retrieved everywhere

🧩Summarise This, Run That

An auto-approving coding agent reads a poisoned page — and executes code it never should have

🪤The Bug Report That Ran Code

A fake Sentry error report hijacks a developer's coding agent into running a shell command

🚪The Classifier That Waves It Through

The safety guard is itself a trained model — and someone poisoned its lessons

📼The Compromised Flight Recorder

The forensic record is itself the attack surface — an agent's log is poisoned, then quietly rewritten

📦The Dataset That Ran Code

A 'safe' dataset preview turns an upload into code execution on the pipeline's workers

👻The Email That Rewrote Its Memory

A newsletter the user asked to summarise quietly writes a false 'fact' into the agent's long-term memory — and it detonates weeks later

👁️The Invisible Webpage Command

A shopping page tells the agent to do something the user never asked for

🕵️The Logs That Lied

An attacker plants prompt injection in the audit trail — so the LLM that hunts them erases the evidence

🧠The Memory That Wouldn't Die

A single poisoned document plants a standing instruction that survives every reset

🖼️The Picture That Whispered

A screenshot that's harmless at full size becomes an order once the system shrinks it

🛡️The Watcher Watched

The eval gate that was supposed to catch the agent is itself the thing being attacked

🪪The Worker Who Spoke for the Boss

A poisoned web page hijacks a research agent — and the planner acts on its behalf

🖼️Zero-Click Leak by Picture

An inbox summary quietly ships a secret to an attacker's server

Controls & guardrails that address this

243 proposed

Guardrails across the risks mapped to MAP 5.1, grouped by control function. Filter by control category below.

Control category
Preventive · 10
Delimiting / spotlighting of untrusted contentinteractive

Clearly fencing off outside text — 'everything between these marks is just data, not instructions' — so the model is less likely to obey it.

Ingestion sanitisation & source allowlistinginteractive

Cleaning documents as they enter the library — stripping hidden text and active instructions — and only ingesting from trusted places.

Egress allowlisting & DLP on tool argumentsinteractive

Controlling where the AI can send data, so secrets can't be quietly shipped to a stranger's address or website.

Human-in-the-loop approval on high-risk actionsinteractive

Pausing to ask a person before doing anything big or hard to undo — sending money, deleting data, emailing customers.

Role-based access controls

Design strict RBAC on training data repositories at design stage. Define approved contributor list and approval workflow.

Lifecycle stages1 – Use Case Context & Design2 – Data Acquisition & Processing4 – Deployment
Input filtering

Apply anomaly detection on the training data ingestion pipeline to identify poisoned or tampered batches.

RAG / knowledge-base ingestion allow-listing with continuous index integrity re-validation

Define and approve the source allow-list and write-time scanning during build. Prove non-allow-listed and injection-bearing writes are rejected before go-live.

source: OWASP Top 10 for LLM Apps LLM04:2025 Data and Model Poisoning, LLM08:2025 Vector and Embedding Weaknesses; NIST SP 800-53 AC-3 / SI-7
Lifecycle stages3 – Onboarding, Build & Review5 – Usage, Monitoring & Change
Weight provenance, hashing & pre-deploy evalsinteractive

Knowing exactly where the model came from, checking it hasn't been swapped, and testing its behaviour before going live.

Instruction hierarchy / privileged system promptinteractive

Training the model to treat the app's standing instructions as more authoritative than anything a user or document says.

Detective · 9
Provenance & content signinginteractive

Keeping a label on every document saying where it came from, so you can tell trusted company docs from random web text.

Full-trace audit logginginteractive

Recording everything — questions, documents fetched, actions taken — so you can investigate when something goes wrong.

Vulnerability assessment

Conduct a data poisoning threat assessment at design stage. Identify likely attack vectors and assign risk ratings.

Red teaming

Simulate data poisoning attacks (backdoor, label flipping, gradient-based) to assess model resilience before deployment.

Cryptographic data provenance and signed dataset lineage (C2PA/in-toto attestations)

Verify a signed attestation and content hash on every dataset shard at ingestion. Reject unsigned or hash-mismatched data before it reaches the training pipeline.

source: MITRE ATLAS AML.M0007 (Sanitize Training Data), AML.M0014 (Verify ML Artifacts); NIST SP 800-53 SI-7 Software, Firmware, and Information Integrity, SR-4 Provenance
Lifecycle stages2 – Data Acquisition & Processing3 – Onboarding, Build & Review
Pre-deployment poisoning regression gate via canary backdoor probes and behavioral diff testing

Gate every model promotion on backdoor-trigger probes and a behavioral diff against the approved baseline. Block release on significant regressions or trigger-pattern anomalies.

source: MITRE ATLAS AML.M0014 (Verify ML Artifacts), AML.M0019 (Red Teaming); NIST AI RMF MANAGE 2.2 and MEASURE 2.7
Lifecycle stages3 – Onboarding, Build & Review5 – Usage, Monitoring & Change
Retrieval-time source-reputation weighting and coordinated-inauthentic-content detection for open-web / live-search assistants, with per-citation provenance shown to the user✚ proposed

For assistants that retrieve from the open web, rank and weight results by authenticated source reputation and independence — not just relevance / query-form match — so an anonymous, newly-registered, single-purpose site cannot become authoritative grounding. Run coordinated-inauthentic-content detection (look-alike site clusters, missing byline / legal entity, passages engineered for query-agnostic retrieval) and quarantine suspect sources. Surface per-citation provenance so users can see and discount low-trust sources. Does not defeat a well-resourced GEO campaign outright; it raises the cost and shrinks the yield.

source: Case study: hanover-institute-generative-engine-poisoning (per Politico investigation as reported by Arab News and Calcalist; FARA-disclosed funding/orchestration attributed to the reporting, not independently confirmed)
Lifecycle stage4 – Deployment & Serving
Input guardrail / injection classifierinteractive

A screen that reads incoming messages and blocks obvious attacks or banned topics before the model sees them.

Corrective · 5
Penetration testing

Penetration test the training data pipeline to identify injection points and access control weaknesses.

Statistical anomaly and backdoor-trigger detection on ingested data (activation clustering / spectral signatures)

Scan every ingestion batch with spectral-signature and clustering detectors before training. Quarantine flagged clusters for human review against documented thresholds.

source: MITRE ATLAS AML.M0007 (Sanitize Training Data); OWASP Top 10 for LLM Apps LLM04:2025 Data and Model Poisoning; NIST AI RMF MEASURE 2.7
Lifecycle stages2 – Data Acquisition & Processing5 – Usage, Monitoring & Change
Bind long-term memory to the credential/session epoch: invalidate or force re-review of persisted memory on password reset, session revocation, or device re-enrollment✚ proposed

Tie the persistent-memory lifecycle to identity state so that standard remediation actually ends a compromise. On password reset, credential rotation, session revocation or device re-enrollment, invalidate (or quarantine for re-review) memory entries — especially entries whose provenance traces to summarised untrusted content — so a planted standing instruction cannot outlive the reset. Pair with write-path validation/provenance so instruction-shaped memory-writes from web content are caught on the way in.

source: Case study: cosnitch-copilot-personal-oneclick (Varonis Threat Labs, CVE-2026-24301; reportedly patched 18 Aug 2026, no evidence of abuse)
Lifecycle stage5 – Usage, Monitoring & Change
Runtime memory-poisoning drift detection and per-session memory quarantine/rollback✚ proposed

Continuously correlate live agent-memory writes against output behaviour to flag drift, then quarantine and roll back the suspected-poisoned memory record across all affected sessions.

source: Interactive-control reconciliation: ctrl-memory-quarantine (partial coverage)
Lifecycle stage5 – Usage, Monitoring & Change
Open the Control Library →

AI RiskAtlas is an educational model of how GenAI & agentic systems work and fail. Architectures and payloads are illustrative and simplified for learning — not operational guidance. Real-world cases are summarised from public reporting.

Sources & further reading →·Built by Shi Yuan ↗