Adopt AI boldly, then close the gap between how fast you adopt it and how fast you can control it.
I lead Accenture's Secure, Responsible AI and Data Protection practice across the Americas: the United States, Canada, and Latin America. The work runs on three fronts. Security for AI puts the governance, guardrails, and evaluation in place so the business can use GenAI and agents without inheriting unmanaged risk. AI for Security rebuilds the security operation around AI because the adversary already has. Data Protection shrinks the blast radius when something gets through anyway.
What I bring to it is the operator's view. I have sat in the CIO chair during a technology shift and know that the danger is rarely the technology itself. It is the gap between adoption and control. The interactive labs on this site exist so that executives can feel that gap rather than read about it: break a hardened assistant, run an incident from the executive seat, red-team your own AI policy.
Adopt AI boldly, with a foundation built to contain its risks.
GenAI and frontier models are reshaping how the business operates, but they also open a new attack surface: prompt injection, model and data poisoning, sensitive-data leakage, and agents that act with real privileges. I help organizations put the foundation, guardrails, and controls in place, including AI governance, model and pipeline security, runtime guardrails, and continuous evaluation, so the enterprise can capture AI's upside without inheriting unmanaged risk.
Attackers already use AI. Your defense has to answer in kind.
AI has lowered the barrier to entry for attackers and collapsed the time from intent to impact. Defending at that speed means rebuilding security operations around AI rather than bolting it on. I help organizations redefine the methodology and processes so AI accelerates detection, triage, investigation, and response, through agentic SOC workflows, automated threat analysis, and human-in-the-loop decisioning that keeps analysts focused on what matters.
You can't always prevent the breach. You can shrink the blast radius.
Every AI and digital initiative ultimately runs on data, which makes data both the prize and the liability. I help organizations minimize exposure across the data lifecycle with the full set of data protection disciplines: data security and classification, encryption, tokenization, and minimization. The goal is simple: reduce how much sensitive data is exposed, and contain the damage if an attacker ever gets through.
Independent · 2025 to present
A multi-tenant Slack alternative in production at cinchme.app, with tenant isolation proven in the database, a browser suite run against the production image, and a release gate that refuses to promote traffic without evidence.
North Highland · 2023 to 2026
A product development methodology built around AI-assisted engineering that reduced cost and time to market by more than 90%.
Accenture · 2026 to present
Practice leadership across the US, Canada, and Latin America, helping enterprises adopt AI without outrunning their ability to secure it.
Adoption is outrunning governance by a wide margin. The defining risk of this era is not an exotic zero-day. It is the debt between how fast we deploy AI and how slowly we learn to control it.
The framework
The danger isn't the AI. It's the gap between how fast you adopt it and how fast you can control it. See where your organization sits in 90 seconds.
Find your gapTry to jailbreak two AI assistants side by side. One has no defenses, one is hardened. See which layer stops each attack.
Take the executive seat in a live AI security incident. Make the calls, then see how I would have played it.
Paste your AI policy, a vendor's claims, or a model card and get a framework-graded teardown against NIST AI RMF, the EU AI Act, and the OWASP LLM Top 10.
Run a simulated agentic AI exploit step by step, then deploy the controls that break the chain. Grounded in the OWASP agentic threat taxonomy.
Ask Paul, a grounded advisor on secure and responsible AI, plus an AI Readiness Assessment and an Executive Briefing generator.