Please enable JavaScript.
Coggle requires JavaScript to display documents.
AI & ML: Privacy, Security, and Ethics - Coggle Diagram
AI & ML: Privacy, Security, and Ethics
Branch 1: Privacy and Security Challenges
Key Issues
Unwanted Data Scraping & Harvesting (collecting user info without clear consent)
Black-Box Model Opacity (hard to see how AI reaches its decisions)
Lack of User Control & Hidden Privacy Settings
Scenario 1 — Employee Monitoring & Data Collection
Situation: An employer uses AI tracking software to monitor remote workers' keystrokes and webcam feeds to judge productivity.
Options:
Option A: Silently run the tracking tool in the background so workers don’t change how they act.
Option B (My Choice): Be completely transparent with staff—let them know about the monitoring, limit data collection strictly to working hours, and give opt-outs for sensitive data.
Ethical Support: Kantian Ethics (Deontology) — Kant argues that people should be treated as autonomous individuals with inherent worth, not just as tools or productivity metrics to be secretly watched without consent.
Scenario 2 — Healthcare Data Ingestion
Situation: A medical tech startup uses private patient health records to train an AI diagnostic tool.
Options:
Option A: Feed full patient records into the AI to make it as accurate as possible, even if it risks exposing private identities.
Option B (My Choice): Anonymize all patient data before training the model and require patients to explicitly opt-in first.
Ethical Support: Rights-Based Ethics — This framework focuses on respecting individual rights, meaning patients have a fundamental right to medical privacy and control over who accesses their health data.
Branch 2: Cybersecurity Advantages
Key Issues
Spotting Anomalies & Network Attacks in Real Time
Automated Threat Hunting & Fast Incident Response
Predicting Malware Behavior & Catching Attack Patterns
Scenario 1 — Automated Threat Response vs. False Positives
Situation: An AI security system detects weird network traffic in a hospital and can automatically block the IP, but it might accidentally lock doctors out of critical patient files.
Options:
Option A: Let the AI automatically shut down suspicious connections on its own without waiting for human approval.
Option B (My Choice): Set the AI to temporarily isolate the threat while immediately alerting a human security analyst to double-check and verify.
Ethical Support: Rule Utilitarianism — Following a strict rule that requires human verification prevents accidental, disastrous shutdowns in critical environments like hospitals, creating the best safety outcome overall.
Scenario 2 — Defensive AI Vulnerability Scanning
Situation: A company uses AI security scanning and finds a major zero-day security flaw in a third-party software vendor connected to their network.
Options:
Option A: Post the security flaw publicly right away to force the vendor to fix it fast.
Option B (My Choice): Practice Coordinated Vulnerability Disclosure—privately report the flaw to the vendor first and give them a reasonable deadline to patch it before going public.
Ethical Support: Social Contract Theory — This maintains implicit trust and fair dealing between security researchers, software vendors, and everyday users.
Branch 3: Cybersecurity Risks
Key Issues
Adversarial ML & Data Poisoning (corrupting model training)
Deepfakes & Hyper-Realistic Phishing Attacks
Model Theft & IP Exploitation through APIs
Scenario 1 — Deepfake Social Engineering Attack
Situation: Scammers use an AI voice-cloning tool to fake an executive's voice over the phone, demanding an urgent money transfer.
Options:
Option A: Rely entirely on an automated AI voice detector to catch fake calls.
Option B (My Choice): Use a defense-in-depth policy that requires human secondary confirmation (like a multi-factor callback) for any wire transfer, regardless of what AI flags.
Ethical Support: Virtue Ethics — Focuses on building good habits and character traits like prudence, caution, and sound judgment instead of relying blindly on technology.
Scenario 2 — Poisoned Training Data in Email Filters
Situation: Attackers deliberately feed bad data into an organization's spam filter so malicious phishing emails sneak through unnoticed.
Options:
Option A: Keep the filter running quietly while trying to fix the code in the background.
Option B (My Choice): Roll the model back to a trusted backup, warn users to be extra vigilant, and re-train the AI using verified data.
Ethical Support: Ethics of Care — Prioritizes looking out for the safety and well-being of the user community by taking proactive steps to protect them from harm.
Branch 4: Ethical Challenges
Key Issues
Algorithmic Bias & Hidden Proxy Variables (like zip codes reintroducing bias)
Corporate Deflection ("Consumers can figure it out on their own")
Job Displacement & Loss of Human Oversight
Scenario 1 — AI Hiring & Resume Screening
Situation: A company uses an automated AI tool to screen resumes, but the model starts favoring male candidates because it was trained on older, biased hiring data.
Options:
Option A: Keep using the software because it speeds up hiring, and try to tweak the code later.
Option B (My Choice): Pause the automated scoring immediately, audit the training data for hidden biases, and make sure human recruiters review all candidates.
Ethical Support: Rawlsian Justice (Justice as Fairness) — Focuses on ensuring equal opportunity and fairness by actively removing systemic bias from the process.
Scenario 2 — Autonomous Systems & Crisis Decisions
Situation: A self-driving car's AI faces an unavoidable crash and has to split-second choose between swerving into a barrier (risking the passenger) or striking a pedestrian.
Options:
Option A: Program the vehicle to always prioritize the safety of the car's owner no matter what.
Option B (My Choice): Program the vehicle using transparent, standard safety rules focused on hard braking and minimizing overall impact without trying to value one life over another.
Ethical Support: Act Utilitarianism — Evaluates the immediate action to reduce net harm and preserve life as fairly as possible in that specific moment.