HAL: Who Signed Off on This?

Christopher Sowa

Every AI decision inside a company eventually lands on someone's desk with a pen. HAL: Who Signed Off on This? is a series of short conversations with the people who sign: how they decide what to approve, where they insist a human still looks, and what they'd need to see before they'd trust a machine with more. (HAL is the computer in 2001: A Space Odyssey, the AI that went rogue. The show is about the people who make sure that never happens on their watch.)

Episodes

  1. 1d ago ·  Video

    Big AI Decisions in Government: Public Safety, Privacy and AI Agents with Fargo CISO Nick Lindhag and Chris Sowa

    Who makes the big AI decisions inside a city government, and how do they keep the public's trust? Nick Lindhag is Chief Information Security Officer and Deputy CIO for the City of Fargo, North Dakota, where he has spent sixteen years supporting departments from public safety to public health. He joins host Chris Sowa to talk about protecting citizen data across 24 departments, why AI policy starts with the data, where a person has to stay in the loop, and why he would repurpose people before cutting them. In this episode: •   What a city IT and security team protects: 24 departments that run like 24 small businesses •   What keeps him up at night: citizen data, from criminal justice records to public health •   How small towns without an IT person stretch their reach with intelligence sharing groups such as InfraGard •   The first question for any AI idea: what problem are we solving? •   Cameras in public safety: transparency, auditing and listening to citizens' concerns •   "We don't want to be a surveillance state, but we also shouldn't reject these tools when they have good purpose" •   Guiding staff on AI tools and agents: education first, blocked by default, opened for a good use case •   If an AI agent moved $100,000 on its own: no victim shaming, and data owners stay responsible •   The seat belt analogy for safeguards •   Confirmation bias and the Google effect: why a person stays in the loop and time saved goes back into auditing •   Self starters in the departments: a separate environment, templates and developer review for staff who build with AI •   Small wins: inventory checkout, big reports and grant writing •   On headcount: repurpose people and keep their experience   Chapters: 00:00  Introduction 00:17  Meet Nick Lindhag, City of Fargo 00:35  Lessons from public safety technology 01:50  What a city IT and security team protects 03:26  How small governments stretch limited IT resources 05:32  How an AI idea gets evaluated 06:43  Cameras, privacy and public trust 09:15  AI agents and guidance for staff 11:59  If an AI agent moved $100,000: who is responsible? 14:49  Accountability, confirmation bias and the human in the loop 17:28  The Google effect: check the sources 19:25  Building an AI culture: supporting the self starters 22:08  Small AI wins in local government 23:30  AI and headcount: repurpose, don't cut 26:37  Close   About the guest: Nick Lindhag is Chief Information Security Officer and Deputy CIO for the City of Fargo, North Dakota. He has been with the city for sixteen years, coming up through networks and servers, and has served in his current leadership role for the last five. About the show: HAL: Who Signed Off on This? is a series of short conversations with the people who sign for AI: how they decide what to approve, where they insist a human still looks, and what they would need to see before trusting a machine with more. Hosted by Chris Sowa, Co-Founder of RedactSure, an AI controls, governance and privacy company. If you sign for AI decisions in your organization and have an answer to "who signed off on this," or are still working one out, write to chris.sowa@redactsure.com. Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/id6811139865 Listen on Spotify: https://open.spotify.com/show/5C2zJHe2qu24q972sSHreE The views expressed are the guest's own.

  2. 2d ago ·  Video

    AI in Insurance: From RPA to AI Agents, Governance and Agentic SDLC with Tom Hardwick of SECURA

    How does an insurance carrier move from RPA to AI agents without losing control? Tom Hardwick is Director of Information Technology, Business Application Services at SECURA Insurance, where he started the enterprise AI governance committee and helped build the intelligent automation team. He joins host Chris Sowa to talk about what reasoning agents opened up after RPA, how IT and InfoSec work as partners, why his team now builds more than it buys, and why the goal is to empower people, not reduce headcount. In this episode: •   Why an IT leader earned insurance credentials in underwriting and claims •   Turning IT language into business terms so ideas reach a technical solution faster •   How SECURA takes in and sizes automation ideas, and why InfoSec gets the first look •   What reasoning agents opened up: a backlog RPA could not handle, document processing and knowledge chatbots •   An underwriting chatbot over 1,066 pages of SharePoint knowledge •   Setting boundaries around where agents live and what they can reach •   Why the team favors MCP servers for speed and simplicity, and keeps RPA for screen work •   Build or buy: agentic SDLC that reaches a demo in six to eight weeks •   What SECURA asks of vendors: AI and automation language in the contract and full disclosure •   Tracking tokens and credits each month, and why upskilling comes before the value story for now •   Responsible use policies and approved tools for personal information •   Advice for leaders: "don't just automate to automate," and empower your people   Chapters: 00:00  Introduction 00:26  Meet Tom Hardwick of SECURA Insurance 01:10  Why an IT leader learned underwriting and claims 02:15  IT as a partner, not the party of no 03:03  Evaluating AI opportunities with InfoSec 05:15  From RPA to reasoning agents 07:35  Boundaries around what agents can reach 09:05  MCP servers or agents that navigate screens 10:33  Build or buy: agentic SDLC and vendor vetting 12:50  Keeping keys and code secure 14:20  Azure Foundry and the data center 15:15  Tracking tokens, cost and value 17:06  Independent insurance agents and AI 18:54  Personal information and responsible use 20:24  Advice for leaders 22:03  Close   About the guest: Tom Hardwick is Director of Information Technology, Business Application Services at SECURA Insurance, where he has worked for about nine years in both IT and business roles. He started SECURA's enterprise AI governance committee and helped create its intelligent automation team. About the show: HAL: Who Signed Off on This? is a series of short conversations with the people who sign for AI: how they decide what to approve, where they insist a human still looks, and what they would need to see before trusting a machine with more. Hosted by Chris Sowa, Co-Founder of RedactSure, an AI controls, governance and privacy company. If you sign for AI decisions in your organization and have an answer to "who signed off on this," or are still working one out, write to chris.sowa@redactsure.com. Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/id6811139865 Listen on Spotify: https://open.spotify.com/show/5C2zJHe2qu24q972sSHreE The views expressed are the guest's own.

  3. 3d ago ·  Video

    AI in Operations: Bidding, Accountability, ROI and Human Judgment with CIO Vince Fattore

    Where does AI pay off inside an operating business, and who answers for it when it gets something wrong? Vince Fattore is a technology executive with more than 25 years leading IT across senior living, distribution and infrastructure. Today he advises companies in manufacturing, distribution, hospitality, transportation and logistics, and road construction. He joins host Chris Sowa to talk about how AI reached the executive table, how he picks the process to start with, where a person still makes the call, and why handing work to AI does not hand off accountability. In this episode: •   How AI first reached the executive table: headlines and conferences, and a CIO who was already testing it •   Choosing where to start: why his team picked the bidding process •   Bid preparation that he says went from more than a week to about three hours •   The four calls he keeps with a person: sensitive data, customer commitments, significant spend and safety •   An AI idea his leadership team pulled back: customer targeting that was too narrow and too intrusive •   Who is accountable when AI gets it wrong: "delegating the workload to AI doesn't transfer accountability" •   AI fatigue and missing ROI: what the hype underestimated in training, data, resources and security •   Why governance, data, resources and goals come first, so the result can be measured •   How AI compares with the move to SaaS and the cloud: more AI work on site and on laptops •   New roles, including chief AI officer and dedicated AI security •   Advice for young IT professionals: "there's a job out there for everybody"   Chapters: 00:00  Introduction 00:32  Meet Vince Fattore 01:39  How AI first reached the executive table 04:10  Choosing where to apply AI: the bidding process 07:12  Confidential data and where a person makes the call 10:10  An AI idea the leadership team pulled back 12:37  Who is accountable when AI makes a mistake? 15:24  AI fatigue and missing ROI 19:04  AI compared with the SaaS and cloud shift 21:03  New roles: chief AI officer and AI security 21:50  Advice for young IT professionals 24:31  Close   About the guest: Vince Fattore is a Chief Information Officer and technology executive with more than 25 years leading IT across senior living, distribution and infrastructure. He now consults for companies in manufacturing, distribution, hospitality, transportation and logistics, and road construction. About the show: HAL: Who Signed Off on This? is a series of short conversations with the people who sign for AI: how they decide what to approve, where they insist a human still looks, and what they would need to see before trusting a machine with more. Hosted by Chris Sowa, Co-Founder of RedactSure, an AI controls, governance and privacy company. If you sign for AI decisions in your organization and have an answer to "who signed off on this," or are still working one out, write to chris.sowa@redactsure.com. Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/id6811139865 Listen on Spotify: https://open.spotify.com/show/5C2zJHe2qu24q972sSHreE The views expressed are the guest's own.

  4. 5d ago ·  Video

    Healthcare AI: Patient Data, HIPAA and AI Agents with Rishov Chatterjee and Chris Sowa

    What changes when the AI handling a patient record is no longer a single model but an agent? Rishov Chatterjee is an AI engineer and data scientist at Curology. Before that he built AI for cancer care and value-based care at City of Hope and Integra Connect, working with hospitals, private practices and insurers. He joins host Chris Sowa to talk about how patient data is kept private as AI moves from predictive models to agents, who answers when records leave an organization, and what all of this costs. In this episode: •   Why predictive models in healthcare went stale, and what changed when large language models arrived •   How patient data is deidentified before it is used for modeling •   Who is accountable if an AI agent takes 10,000 patient records out of an organization •   Who owns patient data, and what has to happen before anyone sells it •   Why a voice or a selfie can identify a patient, and how privacy rules differ across the US, Europe and China •   The hidden cost of letting AI agents do engineering work, and why companies are capping model budgets •   Frontier, open source and locally run models: what each one costs •   How AI is changing insurance claim denials and appeals, and what that means for nonprofits and smaller providers •   How Curology uses AI to track skin conditions while keeping patient photos private   Chapters: 00:00  Introduction 01:17  Meet Rishov Chatterjee, AI engineer and data scientist at Curology 03:32  Patient data: from predictive models to large language models 06:45  APIs, Epic and FHIR compared with agents that read screens 08:27  Deidentifying patient data before modeling 12:18  Who answers if an AI agent takes 10,000 patient records 14:01  Who owns patient data and who can sell it 16:40  Voice, images and biometric privacy across the US, Europe and China 20:34  What data scientists were talking about at DataCon LA 22:55  AI agents doing engineering work, and the hidden cost 25:30  Frontier, open source and foreign models: the tradeoffs 29:07  Who gets the best models inside a company 31:05  Insurers, providers and AI in claim denials and appeals 35:14  What Curology is building with AI and privacy 39:23  Close   About the guest: Rishov Chatterjee is an AI engineer and data scientist at Curology, a personalized skincare company. His earlier work covered personalized oncology and value-based care at City of Hope and Integra Connect. About the show: HAL: Who Signed Off on This? is a series of short conversations with the people who sign for AI: how they decide what to approve, where they insist a human still looks, and what they would need to see before trusting a machine with more. Hosted by Chris Sowa, Co-Founder of RedactSure, an AI controls, governance and privacy company. If you sign for AI decisions in your organization and have an answer to "who signed off on this," or are still working one out, write to chris.sowa@redactsure.com. Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/id6811139865 Listen on Spotify: https://open.spotify.com/show/5C2zJHe2qu24q972sSHreE The views expressed are the guest's own and are not legal or medical advice.

  5. 6d ago ·  Video

    AI Agent Security: Identity, Human Review and Slopsquatting with CISO Jim O'Connell & Chris Sowa

    How should a security leader treat AI agents and AI generated code? Jim O'Connell has more than 30 years in IT. He started as an application developer, served as chief architect at the Federal Reserve, and spent more than six years as Chief Information Security Officer at Old Dominion Freight Line. In 2026 he received the Leadership ORBIE at the CarolinaCISO ORBIE Awards. He joins host Chris Sowa to talk about what changes when AI writes the code and agents do the work, how he weighs the risk of an AI project, and why he calls security a team sport. In this episode: •   AI generated code: why "trust but verify" still applies, and why code review cannot slip •   Slopsquatting: how attackers register the software libraries an AI invents •   Three rules for AI agents: give each one its own identity and permissions, limit how many cycles it can run, and keep a person in the loop on financial decisions •   Where AI is showing up in logistics: routing, freight tracking and forward facing cameras •   Working with insurers and brokers when the subject is security and data risk •   Why security cannot be "the land of no," and what happens when it is •   How attacks have changed since AI arrived •   Whether frameworks such as COBIT, ITIL and NIST are keeping pace, and why a framework has to be adapted to the business •   Bringing AI into a security team as a way to widen what analysts can do •   Why security is a team sport: peer groups, CISA and InfraGard, and advice for mid sized companies and school systems   Chapters: 00:00  Introduction 00:34  Meet Jim O'Connell: developer, Federal Reserve chief architect, CISO 01:24  AI generated code, code review and slopsquatting 03:07  Securing AI agents: identities, loop limits and people in the loop 05:38  AI in logistics: routing, tracking and cameras 07:31  Insurers: friend or foe? 09:27  Weighing AI project risk: security cannot be the land of no 11:30  Shadow AI and getting executives to pay attention 14:53  How attacks have changed with AI 16:29  Are COBIT, ITIL and NIST keeping up? 18:26  Bringing AI into a security team without replacing people 19:55  Nation state attackers and security as a team sport 22:05  Advice for mid sized companies and school systems 24:04  Close   About the guest: Jim O'Connell is a Chief Information Security Officer with more than 30 years in IT. He began at Chubb, was chief architect at the Federal Reserve, and served as CISO at Old Dominion Freight Line for more than six years. He received the Leadership ORBIE at the 2026 CarolinaCISO ORBIE Awards. About the show: HAL: Who Signed Off on This? is a series of short conversations with the people who sign for AI: how they decide what to approve, where they insist a human still looks, and what they would need to see before trusting a machine with more. Hosted by Chris Sowa, Co-Founder of RedactSure, an AI controls, governance and privacy company. If you sign for AI decisions in your organization and have an answer to "who signed off on this," or are still working one out, write to chris.sowa@redactsure.com. Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/id6811139865 Listen on Spotify: https://open.spotify.com/show/5C2zJHe2qu24q972sSHreE The views expressed are the guest's own.

  6. Nov 14 ·  Video

    AI in Schools: Student Data Privacy, Shadow AI and Guardrails with Ron Velez

    What should a school district have in place before it brings AI into the classroom? Ron Velez spent his career running technology for schools, most recently as Director of Technology and CIO of the White Plains City School District in New York. He is the author of a new book, AI 2.0 : Time for Artificial Safe Intelligence  (available on Amazon and IngramSpark). He joins host Chris Sowa to talk about what superintendents and technology directors should ask before approving an AI tool, how to bring parents and unions into the decision, and who answers when something goes wrong. In this episode: •   The "frictionless friend": why AI that always agrees with the user worries him, especially with students •   What a superintendent should review before AI comes into the district •   Shadow AI: staff using tools the district has not approved, and why student information is the concern •   How to bring the PTA, the unions and the critics to the table before a contract is signed •   His first two tests for any AI tool: data privacy and access control, including third party vendors •   What changes when AI agents get access to systems, and who decides what they can reach •   Synthetic instincts: his case for building guardrails into AI models from the start •   An AI tool he was glad to approve (weapon detection on existing security cameras, with people alerted to act) and the kinds he would not •   Advice for school technology directors: earn a seat at the table •   Accountability: policies for staff and a code of conduct for students   Chapters: 00:00  Introduction 00:38  Meet Ron Velez, former CIO of the White Plains City School District 01:31  The frictionless friend problem 02:33  What superintendents should review first 04:59  When students and parents arrive with chatbot answers 07:58  Shadow AI and student information 10:39  Bringing the PTA and unions to the table 14:55  Ron's top tests for an AI tool: data privacy and access control 17:40  What AI agents can see, and the controls around them 20:07  Synthetic instincts: guardrails built into the model 22:32  Lightning round: tools he approved and tools he would not 25:06  Advice for school technology directors 27:34  Who is accountable when AI goes wrong 30:01  The book: AI 2.0: Time for Artificial Safe Intelligence 31:21  Close

  7. Oct 10 ·  Video

    Healthcare AI Governance: Who Owns the Data? Jim Sinkoff, Sun River Health

    Jim Sinkoff of Sun River Health on healthcare AI governance, data ownership, and who answers when it goes wrong. Sun River serves 240,000 patients across 50 sites in Greater New York, and Jim also runs Solutions for Community Health, a company that deploys AI for community clinics nationwide, so he sits on both sides of the table as buyer and seller. His starting point for AI governance: if your organization is the one that gets sued after a breach, then it's your data. Everything follows from that. Contracts that say so. Bi-directional feeds so your data comes back. Verification before any tool goes live.  Also in this episode: Who is responsible if an AI moves $100,000: the controller, the SVP of finance, the CFO and the bank, with the CIO on the second line. Not just the IT guy. What AI can read on staff screens: administratively, everything, once a copilot is launched. Clinically, very little, and why the API layer around the EHR is the point to watch. Robots outside the EHR: how Sun River automates claims billing without letting the automation live inside the clinical record. The denial arms race between payers and providers, and the regulatory circuit breakers that still limit it. "All in on AI" with three conditions: policy, security, and co-creation rather than off-the-shelf. Why he still writes longhand, types his own email, and thinks AI is a new species standing next to us. HAL: Who Signed Off On This? is hosted by Chris Sowa. To be a guest: https://calendly.com/chris-sowa-redactsure/30min

  8. Oct 8 ·  Video

    Need to Know: AI and Student Records, with Dr. Judy White

    Dr. Judy White ran Riverside County's schools, 430,000 students, after 42 years as a counselor, principal and district superintendent across the Inland Empire. Her rule for student files is simple: if it's not needed to know, take it out. The question in 2026 is whether that rule can survive AI. Everything about a student that once sat in separate file cabinets now lives in one record: academics, health, discipline, what's happening at home. That record is what makes AI useful to an overloaded teacher, and what makes it something to handle with care. Her word for where school leaders stand is "quandary." Dr. White walks through the decisions that still belong to a person. AI can surface a pattern of referrals going back years. It will not tell you what is going on in that student's life outside school. The superintendent remains the seat of appeal, and the real question is whether the district noticed the child with his head on the desk before anything escalated. Also in this episode: the team-around-one-student model and why what they learn stays in the room; districts forced to buy their own data back from ransomware; why teachers aren't getting enough AI training and what it costs to fix; social media policy as the rehearsal for AI policy; and the classroom where students found the hallucinations themselves and wrote their own rules. Dr. White now leads MOSAIC, a nonprofit for racial healing and student scholarships, and serves with Think Transcend, which develops superintendents through the Transcendent Leader Institute (thinktranscend.org). HAL: Who Signed Off On This? is hosted by Chris Sowa. To be a guest: https://calendly.com/chris-sowa-redactsure/30min

  9. Sep 24 ·  Video

    AI in College Athletics: Governance, Athlete Data & Security with Tom Livers

    AI is already changing college athletics—but what happens when it moves beyond game film and starts touching athlete performance data, medical information, recruiting, ticketing and operational systems? In this episode of HAL: Who Signed Off on This?, host Chris Sowa talks with Tom Livers, Associate Athletic Director for IT Services at the University of North Carolina at Chapel Hill, about what AI looks like from inside a major college athletic department. HAL_Chris_Tom Tom's team supports far more than computers. UNC Athletics runs roughly 280 events per year, including football games with more than 50,000 attendees and basketball games with more than 21,000. His team helps support ticketing, public safety, concessions, merchandising, broadcast systems, replay technology and the infrastructure behind the fan experience. HAL_Chris_Tom The conversation explores how AI could change that environment, including: AI-powered video analysis that can identify plays, players and game events almost instantly Combining data from athlete wearables to analyze performance and potentially identify injury risk Where athlete data crosses into sensitive educational or medical information How FERPA, privacy offices and data-governance teams affect what new technology can access Why public roster information is very different from private performance or health information The risk of sensitive athletic information creating an advantage for competitors—or even betting markets Why some critical systems, such as ticketing, are intentionally isolated from broader internet access What happens when technology fails in the middle of a live sporting event Why completely banning AI inside an organization may simply push employees to use consumer AI tools outside the organization's controls Tom describes UNC's approach as a more practical middle ground: establish governance, determine which tools are permitted, and keep institutional data inside appropriately controlled environments rather than assuming employees simply will not use AI. HAL_Chris_Tom He also sees significant upside. AI-based video analysis could reduce hours of manual game-film work to seconds and eventually may be able to identify movement patterns associated with injury risk. HAL_Chris_Tom The central question remains the one behind every episode of HAL: When AI gets access to the data and starts influencing the decision, who signed off on it? Guest: Tom Livers, Associate Athletic Director for IT Services, University of North Carolina at Chapel Hill Host: Chris Sowa Podcast: HAL: Who Signed Off on This?

  10. Sep 23 ·  Video

    The AI Delegation Frontier

    How much authority should enterprises delegate to AI—and where should humans remain firmly in control? In this episode of HAL: Who Signed Off on This?, host Chris Sowa speaks with Harry Ghuman, CEO of Alpha Decisions and a former senior executive at Oracle and Palo Alto Networks, about moving beyond AI experimentation toward measurable business value. HAL_Chris_Harry Harry argues that an enterprise is ultimately a collection of decisions and as AI agents become capable of making and executing more of those decisions, organizations need to define what authority AI actually has. The challenge is not only cybersecurity; it is governance, accountability and control. Chris and Harry discuss: Why hundreds of AI pilots may still produce little measurable business value How to identify the business processes and bottlenecks where AI can create real economic impact Why AI authority should be designed around specific processes and decisions The difference between experimentation and consequential AI deployment Why "human in the loop" is not a complete governance strategy Harry's concept of the AI delegation frontier—what organizations should delegate to AI and what they should not Why human judgment remains especially important for highly consequential decisions The risk of eliminating entry-level roles that develop tomorrow's experienced decision-makers The central question is no longer simply, "Can AI do this?" It is: "Who gave it the authority—and should we?" Harry describes the delegation frontier as determining what can safely be delegated to agents while reserving the most consequential decisions for human judgment.

About

Every AI decision inside a company eventually lands on someone's desk with a pen. HAL: Who Signed Off on This? is a series of short conversations with the people who sign: how they decide what to approve, where they insist a human still looks, and what they'd need to see before they'd trust a machine with more. (HAL is the computer in 2001: A Space Odyssey, the AI that went rogue. The show is about the people who make sure that never happens on their watch.)