Ethics, Privacy, Safety, and Governance

Ethical artificial intelligence

Using AI in ways that support fairness, transparency, accountability, and safety.

AI ethics

Study of moral issues related to AI development and use.

Ethical guardrail

A safeguard that reduces harm, such as anonymizing confidential data before using AI tools.

Anonymize confidential data

Remove or mask identifying/private information before sharing data with an AI system.

Validate accuracy

Check whether AI output is correct; useful because generated content can be wrong.

ChatGPT-generated content risk

Generated content may lack common sense, contain false information, or sound confident while wrong.

Common sense

Basic real-world reasoning that a model may fail to apply consistently.

De-identification

Removing identifying information from data; helpful but not always enough by itself.

Re-identification

Figuring out a person’s identity from supposedly anonymous/de-identified data.

Linking databases

Combining datasets, which can increase the risk of re-identifying individuals.

Differential privacy

A mathematical privacy framework that limits how much one person’s data can affect analysis results.

Aggregate querying

Returning summary values from many records instead of exposing individual records.

k-anonymity

A privacy idea where each person’s record is indistinguishable from at least k-1 others.

Federated learning

Training models across devices/locations without directly exchanging raw data.

Secure aggregation

Combining values from multiple parties without revealing each party’s raw data.

Predictive policing

Use of algorithms/data to forecast crime risk or support policing decisions; can reinforce discrimination if biased.

Algorithmic discrimination / bias

Unfair outcomes caused by biased data, design, or model behavior.

Biometric recognition

Using physical/behavioral traits such as face, voice, or fingerprints to identify people.

Safety-critical application

An application where failure can cause serious harm, such as self-driving cars.

Cybersecurity

Protecting systems, networks, and data from unauthorized access or attack.

Surveillance

Monitoring people or environments, often raising privacy and civil-liberty concerns.

Data privacy

Practices and rules that protect personal data.

Data security

Measures that protect data from unauthorized access, corruption, theft, or damage.

Algorithmic transparency

Making algorithmic decision processes and criteria understandable to stakeholders.

Accountability and governance

Assigning responsibility and using frameworks to ensure ethical, legal, and compliant AI use.

Regulatory standards

Rules or official guidelines organizations must follow.