AI Glossary
The terms that decide how AI is built, governed, secured, insured and financed, defined in plain English. Each entry explains why the term matters and links to the latest Laurenceo Intelligence coverage that uses it.
Enterprise AI and Agents
- AI agent
An AI agent is a system in which an AI model plans and carries out multi-step tasks, using tools such as browsers, code, email or business applications, with some degree of autonomy. Agentic AI is... - Retrieval-augmented generation (RAG)
Retrieval-augmented generation is a design in which an AI model first retrieves relevant passages from trusted documents or databases and then answers using them, rather than relying only on what it...
AI Safety and Incidents
- AI alignment
AI alignment is the work of making AI systems pursue the goals and follow the constraints their developers and users intend, including in situations their designers did not anticipate. - AI incident reporting
AI incident reporting is the practice, voluntary or mandated, of notifying regulators, customers or the public when an AI system causes or nearly causes harm, fails in a significant way or is misused. - Hallucination
A hallucination is an output from an AI model that is fluent and confident but false or unsupported, such as an invented fact, citation or figure. - Model evaluation
Model evaluation is the testing of an AI model's capabilities, reliability and risks, such as accuracy, bias, security and dangerous capabilities, against defined measures before and after release. - Red teaming
Red teaming is structured adversarial testing in which experts try to make an AI system fail, misbehave or be misused, so that weaknesses are found before deployment.
AI Governance and Regulation
- AI audit
An AI audit is an independent examination of an AI system's controls, records and outcomes against defined criteria such as a law, a standard or an internal policy. - AI governance
AI governance is the set of policies, roles, controls and oversight an organisation uses to make sure the AI systems it builds or buys are lawful, safe, fair and accountable. It covers who approves... - AI risk management
AI risk management is the process of identifying, measuring, treating and monitoring the risks of specific AI systems, including errors, bias, security weaknesses, privacy harms and over-reliance by... - EU AI Act
The EU AI Act (Regulation (EU) 2024/1689) is the European Union's risk-based law on artificial intelligence. It entered into force on 1 August 2024 and bans certain practices, sets strict... - Explainability
Explainability is the ability to describe, in terms a person can understand, why an AI system produced a particular output, and which inputs and factors drove it. - High-risk AI system
A high-risk AI system, in the EU AI Act, is one used in areas such as credit scoring, employment, essential services, critical infrastructure or law enforcement, or as a safety component of... - Human in the loop
Human in the loop describes AI processes in which a person reviews, approves or can override a system's output before it takes effect. Related terms are human on the loop (monitoring with the... - ISO/IEC 42001
ISO/IEC 42001:2023 is the international standard for an AI management system: the policies, objectives, processes and controls an organisation uses to develop, provide or use AI responsibly.... - NIST AI Risk Management Framework
The NIST AI Risk Management Framework (AI RMF 1.0, published January 2023) is a voluntary US framework for managing AI risk, organised around four functions: Govern, Map, Measure and Manage.
Frontier AI Models
- AI benchmark
An AI benchmark is a standard set of tasks used to measure and compare model performance, for example on coding, mathematics, reasoning or agentic work. - Artificial general intelligence (AGI)
Artificial general intelligence refers to AI systems that match or exceed human ability across most cognitive tasks. There is no single agreed definition or test, and forecasts of when it may arrive... - Frontier AI model
A frontier AI model is one at or near the leading edge of capability at the time, typically the newest large models from a small number of AI laboratories. - General-purpose AI model
A general-purpose AI model is a model trained on broad data that can perform a wide range of tasks and be built into many downstream systems. Large language models are the main example. The EU AI... - Large language model (LLM)
A large language model is an AI model trained on very large amounts of text to predict and generate language. It can answer questions, write, summarise, translate and write code, and increasingly... - Open-weight model
An open-weight model is an AI model whose trained parameters are published so that anyone can download, run and modify it, as opposed to models available only through a provider's service.
AI Investment and Markets
- AI capital expenditure
AI capital expenditure is spending by technology companies and investors on the physical base of AI: data centres, chips, networking and power.
Geopolitics, China and AI
- AI chip export controls
AI chip export controls are government restrictions, led by the United States, on selling advanced AI chips, chip-making equipment and in some cases model weights to certain countries, principally...
Data Centres and Compute
- AI data centre
An AI data centre is a facility built or adapted to run large clusters of AI accelerators, with far higher power density and cooling needs than conventional data centres. - Hyperscaler
A hyperscaler is a cloud provider operating at very large global scale, such as Amazon Web Services, Microsoft Azure and Google Cloud, which supplies much of the world's AI computing capacity.
AI in Bermuda
- Bermuda Monetary Authority (BMA)
The Bermuda Monetary Authority is Bermuda's integrated regulator of financial services, supervising insurers, reinsurers, banks, investment businesses, trust companies and digital asset businesses....
AI in Insurance and Reinsurance
- Bermuda Solvency Capital Requirement (BSCR)
The Bermuda Solvency Capital Requirement is the BMA's risk-based capital framework for insurers, which sets how much capital a company must hold against the risks it carries. - Insurance-linked securities (ILS)
Insurance-linked securities are investments whose returns depend on insurance outcomes, such as catastrophe bonds and collateralised reinsurance. They transfer insurance risk to capital markets... - Parametric insurance
Parametric insurance pays a pre-agreed amount when a measurable trigger is met, such as a wind speed, an earthquake magnitude or an official declaration, instead of reimbursing assessed losses. - Segregated accounts company
A segregated accounts company is a single legal entity whose assets and liabilities are divided into legally separate accounts, so that the obligations of one account cannot be met from another....
Cybersecurity and AI Threats
- CVE
A CVE (Common Vulnerabilities and Exposures) identifier is a unique public reference, such as CVE-2026-12345, assigned to a specific disclosed security flaw so that vendors, defenders and tools can... - Known Exploited Vulnerabilities (KEV) catalog
The Known Exploited Vulnerabilities catalog is a list maintained by the US Cybersecurity and Infrastructure Security Agency (CISA) of software flaws with evidence of active exploitation. US federal... - Prompt injection
Prompt injection is an attack in which malicious instructions are hidden in text, documents or web pages that an AI model reads, causing it to ignore its intended instructions, leak data or take... - Ransomware
Ransomware is malicious software that encrypts or steals an organisation's data and demands payment to restore access or prevent publication. - Software supply chain attack
A software supply chain attack compromises a widely used component, package, update channel or service provider in order to reach the many organisations that depend on it. - Zero-day vulnerability
A zero-day vulnerability is a software flaw that attackers exploit before the vendor has released a fix, leaving defenders with no patch available.
Privacy and Data Protection
- Data protection impact assessment (DPIA)
A data protection impact assessment is a structured review, carried out before a new processing activity begins, of the privacy risks to individuals and the measures that will reduce them. - Personal information
Personal information is any information about an identified or identifiable individual. Under Bermuda's PIPA, organisations that use it must have a lawful basis, use it fairly and keep it secure. - Personal Information Protection Act 2016 (PIPA)
The Personal Information Protection Act 2016 is Bermuda's privacy law. It came fully into force on 1 January 2025 and governs how organisations use personal information, including when AI systems...
Semiconductors and Chips
- GPU (AI accelerator)
A GPU, or graphics processing unit, is a processor designed for massively parallel computation. GPUs and similar AI accelerators are the chips used to train and run modern AI models.
AI in Financial Services
- Model risk management
Model risk management is the discipline banks and insurers use to control the risk of losses from models that are wrong or misused: independent validation, documentation, performance monitoring and...
See also the coverage index and the AI governance research guide.
Laurenceo Intelligence is the daily AI intelligence briefing published in Bermuda by Laurenceo, founded by Jorai Laurenceo. Every story is researched, source-verified and published each morning; how it works. General information only, not legal, regulatory or financial advice.