Building Trust in AI Applications

Most people think having a human approve an AI decision means the decision is safe. It does not. 👀 There is a term for what actually happens when humans rubber stamp AI outputs under time pressure. Automation bias. It is one of the most documented and underreported risks in enterprise AI right now. After 13 years and 200+ deployments, here is what I have learned about building genuine oversight into AI systems. The human reviewing an output needs three things to actually be in the loop. They need to understand what they are reviewing. They need the context to catch what the model gets wrong. And they need to be genuinely empowered to say no without institutional pressure to simply keep moving. Most organisations have none of those three in place. They have a signature process. That is not the same thing. Before any high-stakes AI output reaches a decision point in your organisation, ask these questions. ➡️ Does the person approving this understand the underlying data well enough to catch an error? ➡️ Is there time built in for genuine review or just enough time to click approve? ➡️ What happens if someone says no? Is that genuinely supported? If the answer to any of those is no… you do not have human oversight. You have automation bias with a human signature attached. What does genuine human oversight look like in your organisation right now? #ai #leadership #futureofwork #artificialintelligence #aistrategy #teamhuman #intellectualatrophy #criticalthinking

𝗡𝗮𝗶𝘃𝗲 𝗥𝗔𝗚 𝘄𝗼𝗿𝗸𝘀 𝗶𝗻 𝗮 𝗱𝗲𝗺𝗼. 𝗜𝘁 𝗳𝗮𝗶𝗹𝘀 𝘁𝗵𝗲 𝗺𝗼𝗺𝗲𝗻𝘁 𝗿𝗲𝗮𝗹 𝘂𝘀𝗲𝗿𝘀 𝘀𝗵𝗼𝘄 𝘂𝗽. Embed → retrieve → generate looks clean in a notebook. Real requirements break it: → Questions whose answer is spread across many documents → Industry terms that embeddings get wrong → Bad chunks the pipeline never catches → Answers that live in how things connect, not in any single chunk → PDFs full of tables and images a text-only index cannot read These 5 architectures are how serious teams stay ahead in the agentic AI era: 𝟬𝟭 𝗛𝘆𝗯𝗿𝗶𝗱 𝗥𝗔𝗚 → Dense vectors find meaning. BM25 finds exact words. → Reciprocal Rank Fusion combines both ranked lists. → A safe baseline for almost every team. 𝟬𝟮 𝗚𝗿𝗮𝗽𝗵𝗥𝗔𝗚 → Pull entities and their relationships into a knowledge graph. → Retrieve subgraphs and community summaries, not chunks. → Best when the answer lives in how things connect. 𝟬𝟯 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗥𝗔𝗚 → A planner agent picks the right tool: vector, web, or SQL. → A reasoner agent keeps trying until the answer is solid. → Retrieval becomes a plan, not a single step. 𝟬𝟰 𝗖𝗼𝗿𝗿𝗲𝗰𝘁𝗶𝘃𝗲 𝗥𝗔𝗚 (𝗖𝗥𝗔𝗚) → Grade every retrieval before you trust it. → Correct → answer. Unclear → rewrite the query. Wrong → search the web. → This is what production RAG actually looks like. 𝟬𝟱 𝗠𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗥𝗔𝗚 → One embedding model (CLIP, ColPali) for text, images, and tables. → One vector index. One multimodal LLM. → No more separate pipelines for PDFs with charts. I built a runnable example for each of the five patterns. GitHub link in the first comment. The best teams in 2026 do not pick one. They combine them — hybrid retrieval inside an agentic loop, with a corrective grader, over a multimodal index. Naive RAG is a starting point, not a finish line. That is why most enterprise GenAI projects stall at the demo. Which of these five becomes the default RAG stack in the next 18 months — and which stays a specialized tool?

Two identical CVs. Both written by AI. Both sent to 1,000 people. The only difference: one was named James, one was named Emily. James’s CV got a 97% approval rating. Emily’s got 76% - and reviewers were TWICE as likely to question her competence. Twenty-two percent more likely to question whether she could even be trusted. The feedback on Emily’s CV: “She can’t even write a CV herself - not sure she has the skills to carry out the job.” The feedback on James’s CV: “He just needed a bit of help putting it together.” Same words. Same AI. Different gender. Different verdict. 🚨🚨🚨🚨 How are we STILL HERE?!?!? The study, by former Meta strategist Zehra Chatoo, was reported in Fortune on 10 May. And the most uncomfortable finding wasn’t from older reviewers. It was from Gen Z men. They were 3.5 times more likely to call Emily’s CV “weak.” The generation that is growing up with AI. The generation telling us AI is the great equaliser. The data says otherwise. Chatoo summarised it in a sentence I have not been able to stop thinking about: “When men use AI, we question their effort. When women use AI, we question their integrity.” This is not one study. Harvard Business School has the AI adoption gender gap at 25%. Brookings has found that 86% of the roles with high AI exposure and low capacity to adapt to displacement are held by women. The pattern is consistent and it is widening. The conclusion most people are drawing from this data is “women should be more confident with AI.” I think that misses the point. The bias isn’t in the technology. It is in the people reading the output. Women are not being irrational when they hesitate to use AI openly - they are reading the room accurately. The reputational cost of being seen to use AI is genuinely higher for them. The data confirms what they already sense. The answer is not to ask women to ignore that. The answer is to fix the people doing the judging. To name what is actually happening when an “Emily” CV gets called weak and a “James” CV gets the benefit of the doubt for the same words. To call out the Gen Z men perpetuating a bias they like to claim their generation has moved past. And for women in leadership reading this - use AI anyway. Lead anyway. Document your AI workflows openly. Train your teams in them. Make your usage visible in the rooms where decisions get made. The cost of stepping back from AI in this moment is far higher than the cost of stepping in. We have the data to prove it now. If this resonated, I write about the AI gender gap, ethics, and practical strategy for women in leadership every week in my newsletter. The link is here: https://lnkd.in/emWjxC9t

🤝 How Do We Build Trust Between Humans and Agents? Everyone is talking about AI agents. Autonomous systems that can decide, act, and deliver value at scale. Analysts estimate they could unlock $450B in economic impact by 2028. And yet… Most organizations are still struggling to scale them. Why? Because the challenge isn’t technical. It’s trust. 📉 Trust in AI has plummeted from 43% to just 27%. The paradox: AI’s potential is skyrocketing, while our confidence in it is collapsing. 🔑 So how do we fix it? My research and practice point to clear strategies: Transparency → Agents can’t be black boxes. Users must understand why a decision was made. Human Oversight → Think co-pilot, not unsupervised driver. Strategic oversight keeps AI aligned with values and goals. Gradual Adoption → Earn trust step by step: first verify everything, then verify selectively, and only at maturity allow full autonomy—with checkpoints and audits. Control → Configurable guardrails, real-time intervention, and human handoffs ensure accountability. Monitoring → Dashboards, anomaly detection, and continuous audits keep systems predictable. Culture & Skills → Upskilled teams who see agents as partners, not threats, drive adoption. Done right, this creates what I call Human-Agent Chemistry — the engine of innovation and growth. According to research, the results are measurable: 📈 65% more engagement in high-value tasks 🎨 53% increase in creativity 💡 49% boost in employee satisfaction 👉 The future of agents isn’t about full autonomy. It’s about calibrated trust — a new model where humans provide judgment, empathy, and context, and agents bring speed, precision, and scale. The question is: will leaders treat trust as an afterthought, or as the foundation for the next wave of growth? What do you think — are we moving too fast on autonomy, or too slow on trust? #AI #AIagents #HumanAICollaboration #FutureOfWork #AIethics #ResponsibleAI

This is HUGE (actually)! 🎉 The AI security community just got the blueprint we've been waiting for! The National Institute of Standards and Technology (NIST) has opened public comments on the first-ever AI Security Competency Area for the NICE Framework! As co-lead of the OWASP ML Top 10, I've been working with the community to identify and mitigate AI security risks for years. Seeing NIST now formalise these competencies into a structured framework? This is the validation and roadmap our field desperately needs. AI security is now a core competency domain. The framework describes "foundational knowledge and skills needed to understand AI systems and to use them in a secure manner that maximizes AI's benefits while minimizing potential negative risks." 🧠39 brand new AI-specific knowledge statements (AI-K001 through AI-K039) ⚡7 cutting-edge AI-specific skill statements (AI-S001 through AI-S007) = 🌏46 competency statements covering the complete AI security landscape! What makes me super happy is how this aligns with the incredible work coming out of AU! The Australian Signals Directorate has been absolutely crushing it with their AI security guidance: 🇦🇺 ACSC's AI Security Guidance: "Guidelines for Secure AI System Development" (Nov 2023) "Engaging with Artificial Intelligence" (Jan 2024) "Deploying AI Systems Securely" (April 2024) "AI Data Security Guidance" (May 2025) AI security is finally getting the serious, structured attention it deserves! The AI-K statements are literally your AI security roadmap! : AI-K003: Knowledge of AI bias types AI-K005: Knowledge of AI model vulnerabilities AI-K007: Knowledge of common AI security risks AI-K013: Knowledge of data poisoning cyberattacks AI-K023: Knowledge of misinformation/disinformation vulnerabilities AI-K024: Knowledge of NIST AI Risk Management Framework ⚡ Skills: AI-S002: Skill in developing prompts for generative AI systems AI-S006: Skill in identifying AI hallucinations and mistakes AI-S007: Skill in measuring non-explainable risk This framework brilliantly covers the complete AI security spectrum: 🛡️ Security OF AI: Protecting AI systems from cyberattacks 🚀 Security THROUGH AI: Leveraging AI to supercharge cybersecurity ⚖️ Responsible AI Usage: Ethical, legal, and risk considerations ☢️ Public comment period: June 2 - July 17, 2025 This is our chance to influence how the next generation of cybersecurity professionals will be trained in AI security! Whether you're implementing AI in your org, teaching the next generation, or researching like we do in OWASP - your input matters! From Australia's perspective, this perfectly complements our national AI strategy and ACSC's world-class guidance. We're not just keeping pace with global AI security - we're helping lead it! #AISecurity #Cybersecurity #NIST #AIGovernance #Australia #MLSecurity #ArtificialIntelligence 🔗 https://lnkd.in/gNMwvNks

AI agents are not yet safe for unsupervised use in enterprise environments The German Federal Office for Information Security (BSI) and France’s ANSSI have just released updated guidance on the secure integration of Large Language Models (LLMs). Their key message? Fully autonomous AI systems without human oversight are a security risk and should be avoided. As LLMs evolve into agentic systems capable of autonomous decision-making, the risks grow exponentially. From Prompt Injection attacks to unauthorized data access, the threats are real and increasingly sophisticated. The updated framework introduces Zero Trust principles tailored for LLMs: 1) No implicit trust: every interaction must be verified. 2) Strict authentication & least privilege access – even internal components must earn their permissions. 3) Continuous monitoring – not just outputs, but inputs must be validated and sanitized. 4) Sandboxing & session isolation – to prevent cross-session data leaks and persistent attacks. 5) Human-in-the-loop, i.e., critical decisions must remain under human control. Whether you're deploying chatbots, AI agents, or multimodal LLMs, this guidance is a must-read. It’s not just about compliance but about building trustworthy AI that respects privacy, integrity, and security. Bottom line: AI agents are not yet safe for unsupervised use in enterprise environments. If you're working with LLMs, it's time to rethink your architecture.

I am a Security Engineer at Google with 7+ years of experience. Here are 17 lessons I learned about Threat Modelling working in DevSecOps that made me a better Security Engineer... (It took me a lot of mistakes to learn these, but you don't have to!) 1. Threat modelling starts with the business → if you don’t know what makes money, keeps trust, or keeps systems up, your model is just a diagram, not risk. 2. Draw the system before you “secure” it → users, services, queues, third parties, data stores, and which way data flows; no diagram = fake clarity. 3. Trust boundaries are where real trouble lives → anywhere data or control crosses teams, networks, orgs, or privilege levels deserves extra attention. 4. Model the attackers you actually face → insiders, leaked tokens, overprivileged services, abused workflows are more likely than nation-state zero days. 5. Threat modelling belongs in design docs → if it happens after everything is built, you’re just writing an incident report in advance. 6. Architecture is a security decision → multi-tenant vs single-tenant, shared DB vs per-tenant DB, sync vs async all change which attacks are even possible. 7. Your CI/CD and IaC repos are part of the attack surface → build agents, runners, deployment keys, and pipelines should be on the diagram, not an afterthought. 8. Business logic is where attackers quietly print money → refunds, credits, retries, limits, and edge cases need more modelling than your login page. 9. Good threat models are about assumptions → “only service X can call this API” or “this key never leaves the VPC” should be written down and challenged. 10. A threat model without concrete controls is just a story → each high-risk scenario should end in specific changes to design, config, or process. 11. Prevention without detection is half a job → for every serious threat, ask “how would we know this is happening” and “who gets paged.” 12. You can’t fix everything → be explicit about what you accept, why, and who agreed; unspoken risk is what hurts you later. 13. People and process can undo perfect design → who can approve access, hotfix in prod, change configs, and bypass checks must be part of the model. 14. Complexity hides vulnerabilities → if it takes 20 minutes to explain the data flow, you’re probably missing risks and nobody will maintain the controls. 15. Reuse threat patterns for common flows → login, file upload, webhooks, internal admin tools should have standard risks and standard mitigations you pull from. 16. The best sessions feel like debugging, not a police interview → engineers should walk out feeling “we found landmines together,” not “security blocked us again.” 17. Threat modelling is a habit, not an event → bake a small threat section into every big design and major change; repetition beats a once-a-year workshop. -- 📢 Follow saed ‎for more ♻️ share the insights

Your people don’t fear AI. They fear what you’ll do with it. If you're a leader, let that sink in for a moment. Most employees are optimistic that AI will reduce drudge work and free up time for more meaningful tasks. What they worry about is surveillance, fairness, and being reduced to a data point in an opaque system. That’s not a tooling issue → that’s a trust issue. ⸻ From a leadership coaching lens, this is where you earn or erode trust very quickly: 🔹 If you introduce AI only in the context of cost‑cutting, people will connect the dots. 🔹 If you talk about “augmentation” but never invest in reskilling, people will connect the dots. 🔹 If decisions change and no one can explain why, people will connect those dots too. ⸻ Leaders who navigate this well do three things: ✅ Declare the “red lines”: Be explicit about what AI will not be used for in your company. ✅ Put humans visibly in the loop: Make it clear that people—not models—own the final decisions that affect careers. ✅ Invite challenge: Create safe ways for employees to question AI‑supported decisions and raise concerns. ⸻ Before rolling out any AI initiative that touches people, ask yourself: “If I were on the receiving end of this, what would I need to see, hear, and know to trust it?” Design from that place—and your AI strategy becomes a TRUST strategy, not just a tech strategy. Coaching can help; let's chat. ♻️ Repost it to your network and follow Joshua Miller for more tips on coaching, AI-era leadership, career + mindset. ⸻ #ai #leadership #executivecoaching #culture #mindset #careeradvice #hr

Trust is the real bottleneck to AI impact, not GPUs or models. I went through the SAS Data and AI Impact Report. It is one of the clearest looks at what actually drives outcomes in the enterprise. Here is the short version. You can also find the complete report here – https://lnkd.in/d7XfVKNM What the report highlights • Generative AI usage is up, and agentic AI is rising, but traditional ML still underpins real production work. • Most teams say they “trust” AI, yet many lack the governance, explainability, and monitoring needed to prove it. That gap lowers ROI. • ROI improves when goals are value focused. Customer experience, growth, resilience, and time to value outperform pure cost cutting. • The biggest blockers are weak data foundations, inconsistent governance, and skills gaps. • Maturity varies by industry, but leaders share the same pattern. Centralized data, accountable governance, and an end to end AI lifecycle. Why this helps enterprises • It gives a benchmark. Use trust and impact indices to see where you stand and where to invest next. • It links trust to hard results. Governance is not a checkbox. It is how you improve returns and reduce surprises. • It focuses on foundations. Good data, clear policy, and lifecycle oversight beat ad hoc pilots. My take • Move from “save cost” to “create value.” Prioritize customer experience, decision speed, and new revenue paths. • Treat trust like an operating system. Build a reusable layer for governance, explainability, bias testing, evaluation, and monitoring. Use it across all use cases. • Prepare for agentic AI with data work first. Consolidate data, define permissions, and track lineage. Agents will only be as good as the operating environment you give them. • Invest in skills. Teach builders evaluation and safety. Teach business teams how to measure decision quality. • Start small, measure fast, scale what works. Make ROI reviews a habit, not a milestone. Why this matters now AI has moved from pilots to core workflows. If trust lags, risk scales faster than value. If trust leads, value compounds. This report offers a practical map for leaders to shift from enthusiasm to impact. If you lead data or AI in your company, block time with your team this week. Align on foundations, governance, and near term value. Then execute. #data #ai #agenticai #sas #theravitshow