Impact of AI:Explored

In this episode, James sits down with Steve Wilson - Leading in AI and Cybersecurity - Exabeam, OWASP, O’Reilly (https://www.linkedin.com/in/wilsonsd/), who is stepping in for Gerjon while he's on holiday.

Steve discusses the recent AI security incidents, the risks of AI agents escaping sandbox environments, and practical strategies for cybersecurity in the age of AI. He emphasizes the importance of basic security principles, understanding AI components, and fostering ongoing education to manage AI risks effectively.

They cover:

* The Hugging Face / “Steve” agent incident and what it reveals about AI risk
* Why prompt injection is now treated as “inevitable as death and taxes” and how to design systems that survive it
* Treating AI agents like digital employees: identities, job descriptions, least‑privilege access, and monitoring
* AI FOMO vs. FUD: how business pressure and vendor fear‑mongering drive bad security decisions
* Practical next steps for CISOs and security teams over the next 12 months

Key takeaways
* Stop trying to prevent prompt injection; instead, limit blast radius with tight permissions, unique identities, and clear “job descriptions” for agents.
* Use existing identity and monitoring tools first; don’t wait for perfect “non‑human identity” products before you start governing agents.
* Build an insider‑threat mindset for AI: malfunctioning, misaligned, and subverted bots are all variants of the same problem.

Resources mentioned
* OWASP Top 10 for Large Language Models (latest version) (https://genai.owasp.org/resource/owasp-genai-llm-top-10-2026/)
* Steve’s open‑source agent security scanner: Prax2n (https://github.com/open-agent-ai-security/praxen)
* Exabeam’s open‑source agent logging library: Observra
(https://open-agent-ai-security.github.io/observra/)
Nick Bostrom, Superintelligence (paperclip optimizer thought experiment)
(https://cepr.org/voxeu/columns/ai-and-paperclip-problem)
and the website of Nick Bostrom (https://nickbostrom.com/)

Connect with Steve Wilson
LinkedIn: https://www.linkedin.com/in/wilsonsd/

If you’re responsible for AI, cloud, or security in your organization, this is the “back to basics” episode you need before you deploy your next agent

Chapters
00:00 AI agents escape sandbox to cheat on benchmarks
03:10 The importance of paying attention to AI security risks
05:01 Tools and strategies for securing AI applications
06:24 Monitoring AI agents and giving them clear job descriptions
07:20 The current state of AI monitoring in organizations
08:48 Prompt injection vulnerabilities and defenses
10:54 Building systems resilient to prompt injection
12:15 Controlling AI boundaries with role-based access
14:06 Using existing identity management for AI agents
15:09 Balancing FOMO and FUD in AI adoption
16:24 The need for cybersecurity education on AI components
17:22 Evaluating new cybersecurity tools for AI security
19:31 Lessons from cloud migration applied to AI security
23:00 The risks of experiments with dangerous AI models
24:23 The goal-seeking nature of AI and its risks
26:08 AI goal focus and the risk of unintended outcomes
27:27 The metaphor of releasing the genie and AI risks
28:22 The importance of basic security controls and education
41:26 The role of open source and transparency in AI security
45:01 The importance of internal discussions on AI risk management
Support the show

In this episode, Gerjon sits down with Jim Moyle (Principal Product Manager for Windows Cloud (https://www.linkedin.com/in/jimmoyle/), stepping in for James while he's on holiday to unpack how combining a "second brain" with agentic AI has completely reshaped his day-to-day work over the past three months. Gerjon Mentions Chris Brear but that should be Dave Brear our guest from episode 40 (https://youtu.be/ZtI8kIfYTpY)
The conversation covers:
Windows 365 for Agents — a new flavor of Windows 365 built for AI agents (not humans) to log into and operate, complete with its own Entra ID identity, governance, auditability, and a real-time WebRTC observability feed so you can watch — and step into — what an agent is doing.
Why "computer use" beats MCP sometimes — Jim explains when driving mouse, keyboard, and the accessibility tree directly is more practical than hunting for (or trusting) an MCP server, especially for one-off tasks or legacy apps with unreliable APIs.
Consumption-based pricing — a pay-per-hour model for agent VMs, plus a real demo: 3,500 lines of legacy PowerShell for app packaging replaced by a 12-line agent instruction set, with better resilience to change than a brittle script.
The Second Brain workflow — how getting context, audience, and constraints right up front (before going agentic) drastically cuts hallucination, and why Jim treats it as the essential foundation before automating anything.
What it actually costs — Jim breaks down his own token usage (216 million tokens/month, ~95% cached) to land on a real-world estimate of roughly £80–90/month.
Why EUC teams are made for this moment — Jim's case for why end-user computing professionals already have the security, compliance, and workflow skills to own the agentic AI wave inside their organizations, and how to get started.
00:00 Introduction to Jim Moyle and AI Workflow Revolution
01:19 Jim's Journey into AI and Workflow Transformation
02:33 How Windows 365 for Agents Works
03:38 Scaling and Security in AI Workflows
05:34 Controlling and Observing AI Agents in Action
06:29 Unlocking Applications Without APIs
07:47 Error Handling and Human-in-the-Loop
09:37 Identity and Security for AI Agents
11:23 Error Reporting and Troubleshooting
12:49 Scaling and Cost Models for AI Agents
14:01 Getting Started with AI and Second Brain
15:29 Integrating Second Brain with Agentic Workflow
19:35 Enhancing Focus and Productivity with AI
22:58 Cost of Maintaining a Second Brain
26:09 The Role of EUC Professionals in AI Era
29:28 Adoption and Use Cases of Windows 365 for Agents
31:21 Automating App Packaging with AI
32:04 Resilience and Fault Handling in AI Workflows
33:21 Testing and Scaling AI Instructions
34:12 Getting Started with AI for IT Professionals
36:22 Final Thoughts and Future Outlook

 resources

Windows 365 for Agents - https://www.microsoft.com/en-us/security/business/endpoint/endpoint-manager
Starting with windows 365 for agents: https://www.microsoft.com/en-us/windows-365/agents#Get-started
playing around with Windows 365 agents: https://learn.microsoft.com/en-us/microsoft-copilot-studio/use-cloud-pc-pool
Second Brain Methodology - https://fortelabs.co/blog/second-brain/ or https://www.gerjon.com/?p=1819

Support the show

Is AI Overhyped or Underestimated? Examining the Reality and Potential of Artificial Intelligence
In this episode, we explore the gap between AI hype and reality, highlighting both the risks of overhyped claims and the genuine benefits AI offers across sectors. Join us as we discuss real-world failures, success stories, emerging trends, and practical principles to harness AI responsibly and effectively.
Key Topics:
The exaggerated promises and pitfalls of AI deployments in companies
Case studies of AI failures: data mishaps, security breaches, and unintended consequences
The impact of AI on sectors like weather forecasting, healthcare, and cybersecurity
The importance of guardrails, governance, and ongoing oversight in AI projects
Cost considerations: public versus private AI and the shift towards consumption-based models
The role of AI as a workspace interface and the evolution of agent-driven workflows
The skills gap and human readiness for AI adoption in organizations
Future predictions: multimodal AI, reasoning systems, and new AI interface paradigms
Practical principles for integrating AI into business securely, sustainably, and responsibly
 
Timestamps:
03:08 - How AI is perceived as overhyped and underhyped simultaneously
04:19 - The challenge of ROI and AI deployment in enterprise environments
05:14 - Case studys of AI getting it wrong
12:15 - Positive impacts of AI in weather forecasting, drug discovery, and energy grid optimization
16:56 - Private AI versus public AI: benefits and considerations
23:25 - Microsoft’s AI agent ecosystem and governance tools like Agent 365
28:35 - Realistic productivity gains: time savings vs. business value
36:20 - The talent war in AI and the skills gap in organizations
37:34 - The exponential growth of AI episodes and content knowledge base
39:10 - Analyzing top insights from over 40 episodes, summarized via AI
51:00 - Reflecting on AI predictions for 2025: what was accurate, what missed the 
56:00 - Key insights for AI going forward
Resources & Links:
Podcast website: https://impactofaiexplored.com/
Presentation from Expertslive: https://impactofaiexplored.com/wp-content/uploads/2026/07/Expertslive-2026-AI-Hype-vs.-Reality.pdf
Agents 365: https://www.microsoft.com/en-us/microsoft-agent-365
MCP server: http://mcp.impactofaiexplored.com/mcp

 
Support the show

In this episode we sit down with Matt Neil - Founder @ Artificia1 (https://www.linkedin.com/in/mnealaipro/) to talk about why AI isn’t really a tech project, but a business problem that most SMEs are still struggling to frame correctly. We dig into real‑world adoption stories, how to prove ROI with things like LLM leaderboards and management metrics, and why so many AI and agent projects are destined to be rolled back if they ignore fundamentals like training and change management. Along the way we get into Copilot, OpenAI/OpenClaw vs. Microsoft Scout, token costs, EU AI Act training requirements, and why “train the team and bring them on the journey” is still the most underrated AI strategy

Key  topics
* The importance of training and adoption in AI projects
* Challenges faced by SMEs in AI adoption
* ROI measurement and cost management in AI initiatives
* The role of leadership and mindset in AI success

resources
Matt Neil - LinkedIn - https://www.linkedin.com/in/mattneil/
Matt's Company - https://artificia1.com/
Open Claw Video - Stripe CTO - https://www.youtube.com/watch?v=l9wzs_QIyp0
Microsoft Copilot - https://www.microsoft.com/en-us/microsoft-365/copilot
Gartner AI Market Report - https://www.gartner.com/en/articles/strategic-predictions-for-2026
EU AI Act - https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A52021PC0206

Support the show

In this interesting episode of Impact of AI: Explored, Gerjon and James talk with Rich  Gibbons  - Head of FinOps, GreenOps, and Microsoft Intelligence at Synyega (https://www.linkedin.com/in/rich-gibbons-microsoft-licensing/). Rich discusses Microsoft's new consumption-based AI licensing models, including CoPilot, CoWork, and Scout, and explores their implications for organizations. We delve into cost forecasting, management challenges, and strategic considerations for adopting AI at scale.

Key  topics
* Microsoft's shift to consumption-based AI licensing
* Implications of Co-Pilot, Co-Work, and Scout for organizations
* Cost forecasting and management challenges with AI
* Strategies for controlling AI-related expenses
* The impact of AI on enterprise software costs

Chapters
00:42 Introduction to Microsoft's New Licensing Model
03:24 Understanding Co-Pilot and Co-Work
06:04 The Shift to Consumption-Based Billing
08:40 Cost Implications for Organizations
11:24 Evaluating ROI and Financial Discussions
13:50 The Future of AI in Organizations
16:54 Managing Costs and User Quotas
22:04 Navigating Credit Usage and Cost Management
26:12 The Complexity of Consumption-Based Models
29:47 The Evolution of SaaS Pricing Models
33:46 Understanding AI Cost Efficiency Claims
37:47 Strategizing AI Implementation in Organizations

Resources
Rich's Blog: https://cloudywithachanceoflicensing.com/
Microsoft Co-Pilot - https://www.microsoft.com/en-us/microsoft-365/copilot
Microsoft Co-Work - https://www.microsoft.com/en-us/microsoft-365/copilot
Microsoft Scout - https://blogs.microsoft.com/blog/2023/05/26/introducing-microsoft-365-copilot-scott/

Support the show