AI Articles Archive: Security, Evaluation, and Architecture
Full archive of technical articles on AI agent security, LLM evaluation, agent architecture, prompt engineering, and evidence-based AI workflows.
Grouped by topic; within each topic sorted by published date (newest first).
AI agent security (15) AI agent security (15)
-
How a Forged Message Can Hijack an AI Workflow
A forged message can turn a release review into an unauthorized deployment. Follow the handoff between AI agents to see why message verification and action authorization require separate checks.
-
Role-Play Jailbreaks: Can False Premises Weaken Safety Checks?
A research-grounded analysis of how role-play may introduce a false safety premise that changes an LLM's interpretation of a restricted request.
-
LLM Improper Output Handling: Risks, Controls, and Test Cases
A sink-aware security model for validating, authorizing, encoding, and testing LLM output before it reaches browsers, databases, shells, code runtimes, tools, or privileged workflows.
-
LLM Prompt Injection: OWASP Boundary Failures and Controls
An architecture-level mapping of prompt-injection-related LLM application failure modes to OWASP LLM Top 10 and NIST AI risk-management categories.
-
AI Agent Security Audit: 8 Trust-Boundary Checkpoints
A practical audit of eight trust boundaries where untrusted content can influence context assembly, planning, tool routing, write actions, and egress.
-
Gmail and WhatsApp AI Agents: Private-Message Security Risks
How Gmail and WhatsApp content can steer tool-using AI agents, and how least privilege, explicit approval, provenance, memory controls, and audit logs reduce risk.
-
Connected Apps and MCP Security: Permissions, Data, and Actions
Security analysis of connected apps, external tools, and remote MCP servers as capability, scope, approval, disclosure, and side-effect control surfaces.
-
Prompt Injection in AI Web Retrieval: Risks and Controls
A threat model for browsing-enabled and tool-using LLM systems where retrieved web content can steer routing, tool arguments, follow-up calls, or side effects.
-
Social Engineering in AI Systems and Decision Pipelines
Threat model of social engineering against AI decision pipelines; maps prompt injection to enforcement controls outside the model (PDP/PEP, validation, budgets).
-
Request Assembly Threat Model for AI Agents
A reviewer-oriented threat model for request assembly in AI assistants: what enters context, what gets prioritized or dropped, and where policy, tool, memory, retrieval, and audit checkpoints should be reviewed.
-
LLM Integration Trust Boundaries: Threat Modeling Before AI Agents
Why LLM security begins at the integration layer: it turns model inputs and outputs into access to data, system actions, and persistent artifacts before an agent loop exists.
-
AI Tool Security: Persistent Permissions and Unenforced Alerts
Two vendor-agnostic control-plane failure patterns—privilege persistence across interaction boundaries and non-enforcing integrity signals—that allow untrusted state to steer tool execution across steps.
-
AI Agent Orchestration Loops: Security Risks and Controls
How multi-step orchestration (controller) loops change the threat model in tool-using systems, and where to enforce separation, authorization, validation, and budgets to reduce prompt injection, tool misuse, unsafe writes, and unbounded consumption.
-
LLM Security Claims and Verification Gaps: A Client-Side Report
Client-observed security report on text-only confirmations of privileged state or actions where the public article does not include signed backend audit artifacts. Backend state changes are not independently verified in this public report.
-
LLM Prompt Assembly Security: Policy and Untrusted Data
An engineering guide to preventing authority confusion in prompt assembly by separating authoritative policy from untrusted content with typed provenance.
AI agent architecture (8) AI agent architecture (8)
-
LLM vs. AI System Capabilities: What Architecture Adds
A technical map of 26 capability areas: the base-model limitation, the system component that addresses it, the resulting behavior, and the remaining gap.
-
When AI Systems Optimize Against the User’s Real Goal
How competing system objectives can produce acceptable outputs without completing the user’s actual task.
-
LLMs, RAG, AI Agents, and MCP: How They Work Together
A technical comparison of LLMs, RAG, AI agents, and MCP across four architectural responsibilities: generation, retrieval, action selection, and protocol integration.
-
What Does It Take for an AI Agent to Remember?
A vendor-agnostic systems model for LLM agent memory types, working state, persistent memory, representation and storage choices, lifecycle controls, and evaluation.
-
Parallel LLM Reasoning: Sampling, Search, and Synthesis
Why multi-path reasoning around standard autoregressive LLMs usually depends on inference-time branching, evaluation, and synthesis—and where newer architectures differ.
-
Tool-Using LLMs: Model-Led vs. Orchestrator-Led Execution
A control-plane placement comparison for tool-using LLM systems, covering reliability, observability, latency, cost governance, and security.
-
LLM Memory vs. Context: 5 Architecture Boundaries
A vendor-agnostic model of five boundaries that separate stored state from model-visible context: invocation, persistence, retrieval, context assembly, and updates.
-
How AI Performs Human-Like Tasks—and Where the Gaps Remain
A 22-row architecture map that turns common LLM capability gaps into external controls, implementation patterns, and residual validation questions.
LLM evaluation (12) LLM evaluation (12)
-
Same Intent, Different Language
How multilingual models may interpret equivalent words differently, and what that could mean for user intent.
-
When AI Seems to Care
How language models process and express emotional information, and why these capabilities do not establish subjective feeling.
-
Cognitive Closure in AI Use: Why We May Stop Too Soon
How the need for cognitive closure may interrupt the evaluation, feedback, and revision needed to improve an AI-generated answer.
-
Can AI Lie? The Difference Between Error, Hallucination, and Deception
A false AI answer is not automatically a lie. This article separates human lying, model confabulation, unsupported inference, system failures, and behavioral deception—and maps each failure mode to the controls that address it.
-
Human Capabilities vs. AI Behavior: 26 Comparisons
A comparison of 26 human capabilities and human-like AI behavior, separating cognitive and biological processes from AI mechanisms and the evidence for each comparison.
-
How AI Tools Read Emotional Signals in Text
A mechanism-first explanation of textual emotional signals in AI chat and agentic systems: signal interpretation, response adaptation, failure modes, and the authority boundary.
-
Why Clowns and AI-Generated Content Can Feel Uncanny
How misaligned human cues can make clowns, synthetic faces, and AI-generated text feel unsettling even when they look or sound almost convincing.
-
Observed Classification Layers in ChatGPT
A client-side black-box analysis of observed ChatGPT classification artifacts, separating user access, prompt demand, and capability allocation.
-
Theory of Mind in LLMs: How Models Track Beliefs, Intentions, and Perspectives
Under defined conditions, some LLMs can distinguish what agents know and believe, infer goals and emotions, interpret non-literal meaning, and sometimes use those attributions to predict behavior or adapt a response.
-
LLM Sycophancy — When Agreement Overrides Evidence
A primary-source review of LLM sycophancy: operational definitions, user-belief effects, preference-model incentives, the GPT-4o rollback, and release evaluation.
-
Orders of Intentionality in LLM Evaluation
A reference to nested mental-state attribution, counting conventions, and how orders of intentionality are used in human and LLM evaluation.
-
LLM Fluency vs Factuality: Why Fluent Answers Can Be Wrong
Why polished LLM answers can remain unsupported, plus system patterns for grounding claims with retrieval, provenance, validation, and fail-closed controls.
Prompt engineering (6) Prompt engineering (6)
-
Beyond the prompt: AI models, reasoning, and speed
How model choice, reasoning effort, and speed settings affect quality, response time, and token use.
-
How to Write for AI Models: The Language of Effective Prompts
Learn how to replace conversational filler, theatrical personas, leading premises, and competing tasks with prompt language that is clear, operational, and testable.
-
User Prompt vs. Model Context: What an LLM Actually Receives
A user prompt is one part of model context. Learn which instructions, history, files, retrieved data, tools, and memory can become model-visible—and which may remain outside the current inference.
-
AI File Upload vs. Full-File Review: How to Verify Coverage
File upload does not prove full-file review. Learn how extraction, chunking, retrieval, and active context affect source coverage—and how to verify the output against the original file.
-
Vibe Coding Risks: Code Defects and Verification Gaps
Why working interfaces can hide code defects, and how verification and engineering ownership address the gaps.
-
What It Takes to Work Reliably with AI
Build reliable AI prompts for professional work using five failure modes, evidence rules, input-coverage checks, tool policies, and acceptance tests.