# Fletus Poston III

Fletus Poston III writes and speaks about cybersecurity, with a focus on securing AI systems, model supply chains, and the risks of relying on models and data organizations do not control.

- Profile: https://b2binfluencer.marketing/creators/fletus-poston-iii/
- LinkedIn: https://www.linkedin.com/in/fletusposton
- YouTube: https://www.youtube.com/channel/UCbsODWuR7Di0YhWbGze6pNA
- Topics: artificial-intelligence, supply-chain, cybersecurity, ai-governance, ai-security
- Based on 11 dated posts, latest 2026-07-30

## Summary

Fletus Poston III is most useful for thought leadership on AI security, model supply chain risk, and governance.

- Best for: Cybersecurity vendors seeking a knowledgeable outside voice for practitioner webinars on AI model supply chain security.
- Not for: Hands-on technical enablement, code-level demos, benchmark comparisons, or gated demo assets.
- Check first: Confirm recent audience results, roles, seniority, and booking terms for thought-leadership work.

### Takeaways

- Silent tokenizer updates, dataset poisoning, and runtime drift make third-party model trust risky.
- Reactive patching may not fit AI systems whose behavior keeps changing.
- Governance and operational skills matter for putting AI to work at scale.

## Main ideas

- **Reactive patching will not keep up with AI systems that change continuously**: Poston describes the standard vulnerability cycle — discover a vulnerability, identify the affected systems, deploy a patch, verify the remediation, move on — and says it fails when the thing being defended is AI logic that shifts continuously. The reasoning is about cadence: patching assumes a fixed target and a periodic response, while an AI system's behavior can change between patches. Poston calls continued reliance on that model dangerous. Only the claim and its reasoning appear in the post itself; the argument is developed further in an article the post links to, which is not part of the available material.
- **The third-party model supply chain is where AI risk concentrates**: Poston argues that AI is moving faster than most enterprise security frameworks, and that the most dangerous gap is in the third-party model supply chain. Organizations run critical systems on models they did not build, datasets they did not inspect, and dependencies they do not control, and Poston treats the resulting assumption of trust as exactly what attackers exploit. Rather than staying abstract, the post names specific failure modes and closes on a standard for AI systems: not just powerful, but trustworthy, transparent, and secure — treated as requirements to be built in and verified, not qualities to assume. Poston groups the topic with AI governance, third-party risk, and LLM security rather than presenting it as a purely technical issue.
- **Govern and operationalize AI, don't just understand it**: Poston argues that the real opportunity is not understanding AI but mastering how to govern, secure, and operationalize it at scale. The reasoning is about where value and leverage sit: the people who develop those skills become indispensable at the intersection of cybersecurity, cloud, governance, and artificial intelligence. The post is aimed directly at practitioners, ending on a choice rather than a prediction — the question is not whether AI reshapes your career, but whether you lead the transformation or try to catch up to it.
- **The human factor is part of AI security, not separate from it**: Poston treats people as a first-class part of the AI security problem. The clearest evidence is the subject chosen for a security community talk — the human factor in AI-driven cybersecurity — though the announced topic is all that is available; no slides, recording, or text of the talk were reviewed. The same priority runs through the writing, where the skills around governing and securing AI are framed as what separates practitioners who lead the change from those who fall behind.
- **IT operations is a business function, not just keeping the lights on**: Poston pushes back on the common phrase "keeping the lights on," arguing that strong IT operations teams enable productivity, protect the business, and often serve as the first line of defense for both user experience and security. The priority shows in what Poston hires for: not only troubleshooting ability, but critical thinking, effective communication, and raising the level of the surrounding environment. The post is a hiring notice for an on-site Senior IT Support Specialist in Rock Hill, South Carolina, so it reflects day-to-day operational responsibility alongside the AI security commentary; the text names no employer.

## Work to read

- [AI model supply chain security](https://www.linkedin.com/posts/fletusposton_aisecurity-cybersecurity-supplychainrisk-activity-7456847359889707008-kZtF)
- [Reactive patching will not work with continuous AI logic](https://www.linkedin.com/posts/fletusposton_from-reactive-patching-to-continuous-ai-logic-activity-7486721916985237504-SiFU)
- [Speaking on the human factor in AI-driven cybersecurity (DC704)](https://www.linkedin.com/posts/fletusposton_looking-forward-to-presenting-at-dc704-chapter-activity-7462256877658136576-JPrD)
- [The opportunity is governing and operationalizing AI, not just understanding it](https://www.linkedin.com/posts/fletusposton_the-real-opportunity-isnt-just-understanding-activity-7486495431569420288-VuIw)
- [Hiring a Senior IT Support Specialist (on-site, Rock Hill, SC)](https://www.linkedin.com/posts/fletusposton_we-talk-a-lot-about-keeping-the-lights-on-activity-7466302249040756736-hB9G)

## Brand relationships

- **DC704** (mention)

## Partnerships

No sponsorship or partnership package is documented in the material reviewed. The available posts are Fletus Poston III's own commentary on AI security, model supply-chain risk, and IT operations, plus one hiring notice for a role on that team. None of them states formats, deliverables, pricing, exclusivity, turnaround, or a booking route for brand work. A brand can still read the posts as a signal of the topics available for a thought-leadership collaboration: AI model supply-chain security runs through the material, including a post that links to off-platform commentary on why reactive patching does not hold against continuously changing AI logic, and an earlier post that walks through silent tokenizer updates, dataset poisoning, and runtime drift in third-party models. The clearest sign of public-facing activity is a presentation at a DC704 chapter meeting on the human factor in AI-driven cybersecurity, which shows willingness to speak to practitioner audiences. The reviewed posts do not say whether that appearance was sponsored, what it covered in detail, or what a brand would need to arrange. Treat everything here as a starting point for a direct conversation rather than a published offer.

## Alternatives

- Ian Washburn: Choose Ian Washburn instead when the webinar has to land with university, research-institution, or public-sector security and governance buyers, and when the vendor wants the session framed as institutional trust and resilience rather than product-adjacent threat mechanics. Ian Washburn's own posts describe maturing cybersecurity and governance capabilities in a complex R1 university across vulnerability management, incident response, security operations, awareness and training, zero trust, AI security and governance, data governance, and policy, and argue that security in higher education is an institutional trust, resilience, and governance function rather than only a technical one. The same material says security leaders now have a chance to shape strategy well beyond traditional security operations. A separate post shows the operating scope in detail: recruiting a vulnerability management analyst to run scanning, validation, prioritization, and remediation coordination across endpoint, server, network, identity, cloud, and SaaS, triage events on SIEM and EDR, support incident response, and build automated workflows, dashboards, and metrics. For a vendor selling into research universities or public-sector security programs, that is the buyer's own language and scope, and the governance framing is a closer match to how those institutions actually buy than general AI-security commentary.
- Jeff Williams: Choose Jeff Williams when the webinar should be an argument about how AI changes vulnerability discovery and remediation economics, and when the vendor is willing to share a stage with someone who already has a strong, named product position in application security. The reviewed posts make a specific, arguable case: software stayed relatively safe because vulnerability analysis was slow, not because code was secure or assured, and AI is closing that window so the exposure period collapses. One post uses the image of a coral reef that kept sharks out through sheer complexity, and says the Mythos/Glasswing announcement amounts to an automatic reef mapper whose access will spread. It then tells organizations to update their application security programs for a coming "Dark Software Factory" and to add runtime security at the application layer, naming ADR as a fast way to protect applications even when they have vulnerabilities. The second post goes further, describing an end state where AI streamlines the workflows behind "secure by design" into a software factory that autonomously writes good software and generates a compelling assurance case covering threats, defenses, evidence those defenses work, and runtime proof in production, while noting that people, process, and culture in security resist the change. If the vendor's product is runtime protection or continuous assurance, this creator supplies the most developed argument available in the reviewed material and can pressure-test the session rather than repeat a script.
- Maria C.: Choose Maria C. when the audience is industrial, operational-technology, or critical-infrastructure security teams and the vendor's message is continuity under unpatchable risk rather than enterprise AI model governance. One post opens by saying everyone is talking about how AI helps defenders move faster, then asks the sharper question: what happens when AI enables vulnerabilities to be found faster than organizations can patch them. The answer given is that the future of cybersecurity will be defined less by discovery speed than by how well organizations maintain resilience when vulnerabilities cannot be eliminated immediately, and that resilience-focused protection preserving production continuity will likely become the most important security capability of all. The post then ties that directly to industrial organizations that cannot patch on demand and calls cyber resilience a business requirement, reframing the question from "how quickly can we patch?" to "how do we operate safely while we manage the risk?" The second post shows channel-facing work: organizing a first specialist training in Istanbul with a regional distributor, an instructor, and a room of cybersecurity professionals, and thanking participants, the distributor team, and colleagues. For a vendor whose product protects environments that cannot take a patch window, that is the buyer's daily constraint, and the training post shows the account reaches enablement and channel audiences rather than passive readers.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/