PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteAI can make a cybersecurity team look stronger without showing whether its people can investigate an unfamiliar attack, challenge a bad recommendation or make a defensible decision when the tool is unavailable. That is the talent-crisis risk identified by Tinus Green in his August 10, 2026 commentary for Cybersecurity Insiders: automation may accelerate genuine expertise, but it can also hide missing expertise behind faster output.
The distinction matters because the available figures describe reported attitudes and behavior, not a proven, industry-wide decline in skill caused by AI. Leaders therefore need to measure reasoning, verification and independent performance alongside ticket counts and response times.
As an Amazon Associate I earn from qualifying purchases.
What “AI hiding the cybersecurity skills gap” means
Security work produces many outputs that are easy to count: alerts closed, detections tuned, reports written, indicators enriched and incidents escalated. Generative AI and automated analysis can increase those numbers quickly. A faster result, however, does not reveal whether the analyst understood the evidence, noticed a missing signal or could reproduce the decision without assistance.
Green’s argument is about that visibility problem. If a model supplies a plausible investigation summary, an analyst may complete the task on schedule while learning less about the underlying system or attacker behavior. Under deadline pressure, repeated acceptance of confident output can turn a capability aid into a substitute for judgment. The concern is not that every use of AI reduces skill; it is that ordinary productivity metrics can conceal the difference.
That is an argument about risk, not a causal finding. The percentages cited in Green’s article report what respondents said about productivity, confidence, training and AI use. They do not prove that AI caused anyone’s skills to decline, nor do they establish that every security team is affected in the same way.
What the reported numbers actually say
Green’s article attributes the following cybersecurity-worker results to a survey discussed in his 2026 commentary. The linked TryHackMe landing page identifies Censuswide as the research platform, but the accessible material does not provide the sample size, geography, field dates or exact wording needed to judge representativeness.
| Reported result | What it describes |
|---|---|
| 82% | Cybersecurity workers who said AI made them feel more productive, but not more skilled. |
| 38% | Cybersecurity professionals confident that their current skills would remain valuable in the near future. |
| 18% | Respondents expecting to take on tasks or responsibilities they had not learned to handle. |
| 76% | Respondents who regularly leaned on AI for tasks they did not fully understand. |
| 66% | Respondents who said they would struggle to do their job without AI tools. |
| 98% | Respondents who admitted using AI tools their organization had not approved. |
| 18% | Respondents who said training resources existed but felt disconnected from day-to-day work. |
These are article-reported percentages, not universal measurements of the cybersecurity workforce. The underlying report should be consulted before using them to set a benchmark, forecast staffing needs or compare countries and job levels.
A second set of figures should not be merged with the first
The accessible TryHackMe report landing page displays different results for workers overall: 62% said AI made them more productive but not more skilled, 43% regularly used AI for work they did not fully understand, 40% said they would struggle to perform effectively without AI tools, and 66% said AI made decision-making faster but not smarter. The page does not show a publication date or the methods behind those figures. Because the populations and questions are not established as identical, the overall-worker numbers cannot validate or be combined with Green’s cybersecurity-worker percentages.
Why productivity can rise while capability stays flat
Automation removes practice as well as effort
Skill grows through repeated choices: deciding which signal matters, testing a hypothesis, interpreting an incomplete log and explaining why one response is safer than another. If an AI assistant performs those steps on routine work, a practitioner may gain time but lose opportunities to form a mental model. The loss is usually invisible until a novel incident falls outside the model’s training data or the organization’s usual playbook.
Rank #2
- Matt-laminated and greaseproof pages ensure glare-free reading and long life
- The outside covers are made from a new rubberized material for better Handling and Grip
- All the Tool Holder Identification Sections now include a full INCH section along with a METRIC section
- Updated and Improved Index Searching
Fluent answers can suppress useful doubt
Security decisions often involve ambiguous evidence. A polished explanation can encourage an operator to stop asking what the model assumed, which data it ignored or how an attacker could evade the proposed control. Ashu Savani, TryHackMe’s co-founder, states the danger plainly on the report page: “AI doesn’t know when it’s wrong. That isn’t a flaw they’re still working on, it’s a known limitation. The risk isn’t that employees use AI, but that they’ve stopped being the last line of defense between confident output and costly mistakes.” The quotation appears on the report landing page, whose accessible text does not show a publication date.
Dependency can be mistaken for competence
An analyst who performs well with a copilot may be highly capable, lightly supported or unable to work independently; the output alone cannot distinguish those cases. The 66% figure reported by Green—respondents who would struggle without AI—therefore raises a measurement question rather than proving incompetence. Teams need to test what happens when assistance is removed and when the scenario changes.
Acceleration versus masking: a practical distinction
AI is accelerating expertise when the human operator can state the objective, choose appropriate evidence, recognize uncertainty, verify the recommendation and explain the trade-offs. The tool shortens mechanical work while the practitioner retains control of the reasoning.
AI is masking a gap when the operator cannot explain the conclusion, accepts an answer that conflicts with the evidence, fails to identify a missing data source or cannot continue after the tool is disabled. A useful review asks not only “Was the incident closed?” but also “How did the analyst know this was the right action, and what would change the decision?”
- Explainability: Can the person reconstruct the decision in plain language and identify the assumptions?
- Signal selection: Did they prioritize the indicators that distinguish a real attack from noise?
- Verification: Did they test the AI output against logs, telemetry, documentation or a second method?
- Novelty: Can they reason through an unfamiliar or incomplete-information scenario?
- Resilience: Can they perform the critical steps if the model, data feed or integration is unavailable?
These axes operationalize Green’s recommendations; they are not a validated scoring system or a guarantee of future performance.
How security leaders can measure the hidden gap
Replace a single productivity score with paired assessments that reveal both outcome and reasoning. The following process can be run on selected investigations without removing AI from normal work.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →- Define a consequential task. Choose an alert triage, identity compromise or malware investigation in which a wrong decision has a clear cost. Specify the evidence available and the acceptable time window.
- Capture an independent baseline. Have the practitioner work without generative assistance for a bounded portion of the exercise. Record the signals they selected, hypotheses considered, uncertainty and proposed action.
- Repeat with approved assistance. Allow the normal AI tools and require the practitioner to preserve prompts, outputs, edits and verification steps. Do not treat the model’s answer as the person’s explanation.
- Score the reasoning before the result. Review whether the analyst identified relevant evidence, challenged assumptions, weighed trade-offs and stated what would change the decision. Then score accuracy, completeness and time.
- Introduce a novel variation. Change an indicator, remove a familiar log source or add contradictory evidence. This tests transfer rather than memorization of a playbook.
- Discuss the delta. A large speed gain with stable reasoning is useful acceleration. A large speed gain accompanied by weaker explanations, missed signals or inability to continue independently indicates a training or control problem.
Documenting both assisted and unassisted work also makes performance conversations fairer. A slower analyst who identifies a dangerous assumption may be demonstrating more capability than a faster analyst who accepts an incorrect summary.
Controls for safe, accountable AI use
Make approval and data handling explicit
Green’s article reports that 98% of respondents admitted using AI tools their organization had not approved. That figure should prompt a policy and workflow review, not a blanket assumption about every company. Publish an approved-tool list, define what data may be entered, and provide a sanctioned alternative for common tasks. If staff must bypass a policy to meet a deadline, the policy is not functioning as a control.
Require human sign-off on consequential actions
For containment, access changes, detection suppression and customer notification, record the responsible person, evidence reviewed, AI assistance used and verification performed. The sign-off should state uncertainty and the rollback plan, rather than merely confirming that an AI-generated recommendation was followed.
Sample the work for quality, not only speed
Review a rotating sample of closed cases. Look for unsupported certainty, copied explanations, untested indicators and decisions that do not match the available telemetry. Feed recurring errors into exercises and playbook updates. This creates a feedback loop in which automation reveals where judgment needs strengthening.
Rank #4
Training that matches real AI-augmented work
Green reports that 18% of respondents felt training resources existed but were disconnected from their daily work. A course that teaches concepts in isolation will not necessarily prepare an analyst to question a model during a live investigation. Practice should resemble the information quality, tool constraints and time pressure of the job.
Build scenarios around incomplete information
Exercises should include missing logs, noisy indicators, conflicting timestamps and plausible but wrong AI suggestions. Require learners to state what they know, what they are assuming and what evidence they need next. Include attacker behavior and persistence choices so the exercise develops investigation and adversarial thinking, not just button-clicking.
Grade the process as well as the answer
A useful rubric awards points for selecting relevant signals, testing alternatives, explaining trade-offs, identifying uncertainty and deciding when to escalate. Accuracy remains important, but a correct answer reached through an unsafe assumption should not receive full credit.
Use browser-accessible practice where it fits
TryHackMe describes browser-based hands-on cybersecurity training, business training and an AI Security learning path in its materials. It is an example of a digital training service, not independent evidence that any particular course resolves the workforce problem. Organizations should map any platform or lab to their own telemetry, policies and incident procedures, then verify that learners can transfer the skill to unfamiliar cases. The relevant materials are described at TryHackMe’s report and training resource page.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteWhat the evidence does—and does not—establish
Green is identified as a TryHackMe Senior Content Engineer and MWR CyberSec Head of Consultancy, and his piece, “Automation Is Hiding Cybersecurity’s Real Talent Crisis,” was published by Cybersecurity Insiders on August 10, 2026. Read the article as expert commentary that frames a plausible operational risk, not as a controlled study of skill decline. The source is available at Cybersecurity Insiders.
The linked landing page names Censuswide but does not expose the sample, geography, dates or exact questions for the cybersecurity-worker results. It also shows a different set of overall-worker percentages. Until the underlying methods are available, the figures should not be merged, generalized to all organizations or treated as independently validated estimates of the talent gap.
Green also discusses historical changes such as cloud adoption and employer demand. The linked material does not independently substantiate those historical claims, so they should remain attributed to his framing rather than presented as established trend data.
A decision framework for managers
Before expanding an AI deployment, ask four questions in order:
Recommended Free Tools
- Capability: Which human judgment is the tool supposed to accelerate, and how will you observe that judgment?
- Failure: What happens when the model is wrong, unavailable or given incomplete data?
- Evidence: Which assisted and unassisted exercises will reveal whether reasoning is improving or being bypassed?
- Accountability: Who approves high-impact actions, records verification and owns the outcome?
If a team cannot answer those questions, adding automation may improve throughput while making the real capability level harder to see. If it can answer them, AI becomes a measurable aid: faster routine work, preserved human understanding and clearer evidence of where training is needed.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
