Glossary

A chip built to speed up one kind of heavy computing. In AI, GPUs and TPUs are the main accelerators. Lawrence Berkeley National Laboratory (LBNL) uses “accelerated servers” for servers fitted with GPUs for AI work. 11 Site glossary

A written rule set stating which AI tools and uses are allowed or forbidden for an organization’s staff and contractors, usually tied to data classes and human-review requirements.

A term used in New York City Local Law 144 for software that substantially assists or replaces discretionary decision-making in hiring or promotion and meets the City’s regulatory definition.

An AI system that not only answers but takes multi-step actions, such as opening software or filling forms, under defined permissions.

A proposed future AI system that could match or exceed human performance across most cognitive tasks; not a description of current commercial products.

The European Union’s Regulation (EU) 2024/1689 on artificial intelligence, which phases in obligations by risk category from 2025 through 2028.

AI that takes actions for you (browsing, filling out forms, running tasks over several steps) rather than only answering questions. Site glossary

Computer systems that do tasks we usually associate with human thinking, like understanding language, recognizing images or making predictions. Site glossary

A computer-made person who appears on screen and speaks a script you type, with a matching voice. Companies use them for training and marketing videos, and some services let you make an avatar of yourself. HeyGen and Synthesia are examples. 12 Because avatars can copy a real face and voice, see also deepfake and voice cloning. See AI Systems . Site glossary

A web browser with an AI assistant built in. It can summarize the page you are on, answer questions across your open tabs and try tasks such as filling in forms. Perplexity’s Comet is one; OpenAI retired its Atlas browser in August 2026 and moved those features into ChatGPT. 13 Hidden text on a web page can trick the assistant (see prompt injection), so keep it away from banking and other sensitive accounts. Site glossary

Cooling a data center by blowing cooled air past the equipment. LBNL notes that air-cooled chillers are widely used and are preferred where minimal on-site water use is crucial. 11 Site glossary

A computer or network physically cut off from the internet and from other networks, so data cannot leak out or be reached from outside. Used for very sensitive work. See an example in the news . Site glossary

NIST’s Artificial Intelligence Risk Management Framework (AI 100-1, Jan 2023), a voluntary U.S. framework organized around Govern, Map, Measure, and Manage.

The work of finding the risks of AI, such as bias, data-security problems and attacks, and putting processes in place to avoid or reduce them. 14 Site glossary

Making an AI system behave in line with human values and goals so that it is helpful, safe and reliable. IBM describes it as encoding human values and goals into AI models. 15 Site glossary

The EU AI Act annex listing high-risk AI use cases (for example certain employment, credit, and biometric uses) that trigger stricter duties when applicable.

A defined way for one software system to request services from another, commonly how applications call an AI model.

EU AI Act transparency duties for certain AI interactions and synthetic content, with application dates on the European Commission timeline.

A hypothetical AI with human-level ability across a wide range of tasks. It does not exist today, and researchers do not agree on exactly what would count as AGI. 16 It is a different idea from super intelligence, which would go beyond human ability. Site glossary

A hypothetical AI system that substantially exceeds human cognitive ability across nearly all domains; a research and policy concept, not a deployed product class.

A process where software outputs materially affect a person (credit, hiring, benefits) with limited or no human reconsideration in the workflow.

The tendency of people to over-trust machine recommendations relative to other evidence.

Power that is generated and used on the customer’s side of the electric meter, so it does not pass through the public grid first. A data center built next to a power plant can be described this way. See an example in the news . Site glossary

A standard test used to compare AI models, such as a set of exam-style questions scored the same way for every model. A high score shows how a model did on that test, not how it will do on your own task. Site glossary

When an AI’s results are skewed or unfair to some people or groups, often because the data it learned from reflects existing inequities. IBM and NIST both point to training data as a source of bias. 17 , 1 Site glossary

An independent or documented statistical review of an automated tool’s outcomes across protected groups; required for covered AEDTs under NYC Local Law 144.

Systematic differences in model outputs across groups or contexts that reflect imbalances or gaps in training data, objectives, or deployment.

Coalition for Content Provenance and Authenticity; an open technical standard for Content Credentials that record origin and edit history of digital media.

An open standard for attaching a tamper-evident record to photos, video and audio that shows where they came from and how they were edited. C2PA, the coalition behind it, calls the result Content Credentials and compares it to a nutrition label for digital content. 18 Many files do not carry one, so lacking credentials proves nothing. Site glossary

A way to add up different greenhouse gases as one number: the amount of carbon dioxide that would have the same warming effect. Site glossary

Documented handling of electronic evidence so its integrity can be shown in audits or proceedings.

An app you talk to in plain language, such as ChatGPT, Claude or Gemini, that runs on a large language model. Site glossary

A product interface, usually conversational, built on one or more models that accepts prompts and returns text, images, or other media.

OpenAI’s chatbot. OpenAI offers it on the web and in apps for Windows, Mac, iPhone and Android. 19 See Which AI? for how it compares. Site glossary

A machine that cools water or another fluid, the way an air conditioner does, so the fluid can carry heat away from a data center’s equipment. LBNL describes both air-cooled and water-cooled chillers as widely used in data centers. 11 Site glossary

Cybersecurity and Infrastructure Security Agency, the U.S. federal agency that issues guidance on cyber and critical-infrastructure risks, including AI-related threats.

Anthropic’s chatbot. Anthropic says it works on the web, iOS, Android and desktop. 20 See Which AI? . Site glossary

A data-center cooling design that recirculates coolant rather than continuously evaporating large volumes of fresh water.

A model offered only as a remote service, without publicly releasing weights that others can run independently.

Computing capacity provided over the internet by a third party; prompts and files are typically processed on the provider’s servers.

Renting computing power, storage and software over the internet instead of owning the machines yourself. NIST defines it as on-demand network access to a shared pool of configurable computing resources. 21 Site glossary

An AI tool that takes a programming task, such as “fix this bug,” works through it on its own, and hands back the finished change for a person to review. Google’s Jules is one example; Cursor, Claude Code and Codex offer similar modes. 22 See AI Systems . Site glossary

A data center where companies rent space, power and cooling for their own servers instead of building a facility themselves. LBNL tracks colocation data centers as a separate category from hyperscale and from in-house ones. 11 Site glossary

Short for computing power: the chips, time and electricity used to train and run AI. “More compute” means more chips running for longer, which means more data center capacity. Site glossary

EU AI Act process for demonstrating that a high-risk AI system meets applicable requirements before being placed on the market or put into service.

The same AI company can handle data differently for individual and business accounts. OpenAI, for example, says it may use content from individual services such as ChatGPT to train its models unless you opt out, and by default does not use inputs or outputs from ChatGPT Business, Enterprise, Edu or its API. 23 Site glossary

C2PA-based, cryptographically signed metadata describing how a piece of media was created or edited; not by itself a deepfake detector.

Technical and process methods (for example C2PA content credentials) that record how media was created or edited so recipients can inspect origin claims.

All the text a model can look at when it writes a reply, including the reply itself. Anthropic compares it to the model’s “working memory,” which is different from the data the model was trained on. 24 Site glossary

Microsoft’s AI assistant. Microsoft offers a limited free version and sells paid versions bundled with Microsoft 365 plans. 25 See Which AI? . Site glossary

Whether AI companies may train models on copyrighted books, art, music and news without permission is an open legal question. The U.S. Copyright Office has been studying it and has published a report on generative AI training in pre-publication form, along with separate parts on digital replicas and on whether AI-generated works can be copyrighted. 26 Not legal advice. Site glossary

The general-purpose chip that runs a computer’s everyday tasks. AI work leans on GPUs and other accelerators because they handle many calculations at the same time. Site glossary

Child sexual abuse material. Multiple jurisdictions criminalize AI-generated CSAM alongside traditional CSAM; see also TAKE IT DOWN and state child-safety statutes.

Sexual images or video of children; many statutes also cover AI-generated depictions that appear to show children.

An app for writing software, made by Anysphere. It looks like a regular code editor, but you can describe what you want in plain English and its AI writes, changes or explains the code, or runs an “agent” that works on a coding task by itself. It is built mainly for programmers. There is a free plan; paid plans are $8, $30, $100 and $200 a month, and team plans start at $30 per person a month. Prices as of October 19, 2026. Paid Cursor plans are now sold together with Grok and X features. 27 See AI Systems . Site glossary

AI methods that analyze images or video, such as object detection or face matching.

A building full of computers. AI runs in data centers, which is why AI's electricity use is a growing issue. 7 Site glossary

How long a company keeps what you typed, said or uploaded. It varies by company, plan and settings. See Chat privacy for how to check yours. Site glossary

The principle that stored data is subject to the laws of the jurisdiction where it resides.

Synthetic or heavily altered audio, image, or video made to depict a real person saying or doing something they did not.

A kind of machine learning that uses neural networks with many layers. IBM says deep learning models power most state-of-the-art AI today. 28 Site glossary

Under the EU AI Act, an organization that uses an AI system under its authority, as distinct from the provider that develops or places the system on the market; duties depend on risk tier and role.

A mathematical approach that limits how much an individual’s data can influence published results or models.

EU legislative adjustments affecting some AI Act application dates for Annex III and product-embedded high-risk rules, as described on the Commission AI Act timeline.

Direct water use is water a data center uses on site, mostly for cooling. Indirect water use is water used to generate the electricity it runs on. LBNL estimated U.S. data centers directly consumed about 66 billion liters in 2023 and indirectly nearly 800 billion liters through the electricity they used. 11 See also the myth . Site glossary

Texas Department of Information Resources, which certifies certain state AI and cybersecurity training programs.

Data protection impact assessment under GDPR (and similar laws): a structured review of risks to individuals before certain high-risk processing.

An AI model that can support both civilian and military or cyber-offensive applications; export-control and safety policies sometimes single these out.

Running computing close to where data is created or used, such as on your device or a nearby server, instead of in a faraway data center. Cloudflare says the goal is to reduce delay (latency) and bandwidth use. 29 Site glossary

The network of power plants, high-voltage lines and local wires that delivers electricity to homes and businesses. Large data centers draw their power from it. Site glossary

A numeric representation of text, images, or other data that models use to compare similarity and retrieve related content.

Lists of numbers that represent the meaning of text, so a computer can tell which pieces of text are related. OpenAI says they are commonly used for search, grouping similar items and recommendations. 30 Site glossary

A directive from a president or governor to executive agencies; it does not by itself repeal a statute.

Electric Reliability Council of Texas, which operates most of the Texas power grid.

The European Union’s law on AI, Regulation (EU) 2024/1689. The European Commission calls it the first comprehensive legal framework on AI worldwide and says it sets risk-based rules for AI developers and users. 31 See Rules & jobs . Site glossary

Systematic testing of model behavior on defined tasks, including accuracy, safety, and bias metrics.

Cooling by letting water evaporate, which carries heat away, as in a cooling tower. It consumes water. LBNL notes that the evaporation in cooling towers has raised concerns about data center water use and availability locally. 11 Site glossary

Techniques and disclosures intended to make model behavior or decision factors more understandable to humans.

A U.S. legal rule that lets people use some copyrighted material without permission, decided case by case by courts. The Copyright Office keeps an index of court decisions to help the public follow it. 32 How it applies to AI training is still being worked out. Site glossary

Federal Communications Commission, the U.S. agency that regulates interstate communications.

Training across devices or organizations so raw data stays local while model updates are shared.

Training an existing model a little further on specific examples so it gets better at one job or style. OpenAI describes it as training models to excel at a task. 33 Site glossary

Floating-point operations; a unit used to describe the computational work of training or running a model.

A large, general-purpose model trained on broad data that can be adapted to many different tasks, instead of being built for just one. Site glossary

Informal policy term for the most capable general-purpose models at a given time; some state laws (for example California SB 53) define related covered-model thresholds.

Federal Trade Commission, which enforces certain federal consumer-protection and NCII-removal rules.

Google’s AI assistant. Google offers a free version with paid plans above it. 34 See Which AI? . Site glossary

Google’s research tool that answers questions only from the files you give it, such as PDFs, documents, web links and videos. Each answer comes with numbered citations you can click to see the exact passage it used. Example: upload a lease or a service contract and ask “When can I cancel, and what does it cost?” It points to the clauses so you can check them yourself. It can also turn your files into audio summaries, slide decks, mind maps and flashcards. It was renamed from NotebookLM in July 2026, and the old address still works. It is free with a Google account (up to 50 sources per notebook); more room comes with Google AI plans: Plus about $4.99, Pro about $19.99, Ultra $99.99 or $199.99 a month. Prices as of October 19, 2026. It is not a lawyer: check important clauses with one. 35 , 36 See AI Systems . Site glossary

AI that creates new content (text, images, audio, video or code) instead of only sorting or labeling existing content. ChatGPT and image generators are generative AI. Site glossary

Units of electric power. One megawatt is 1,000 kilowatts, and one gigawatt is 1,000 megawatts. Data center sizes are usually quoted this way. These describe power at one moment, not energy used over time. See an example in the news . Site glossary

General-purpose AI under the EU AI Act: models trained at scale that can be adapted to many downstream tasks, with tiered transparency and evaluation duties.

Under the EU AI Act, models that can perform a wide range of distinct tasks and can be integrated into many systems.

A chip that can do thousands of calculations at the same time. It was first built for video game graphics and is now one of the main chips used to train and run AI. Site glossary

Hooking a new power source, or a very large new customer, up to the electric grid. FERC explains that for new power plants it involves studies of how the connection affects grid reliability and what upgrades it needs. 37 Waiting for these studies is one reason projects take time. Site glossary

An AI assistant made by xAI, available on the web, in phone apps and inside X. 38 See Which AI? . Site glossary

Connecting model outputs to retrieved documents or tools so answers can cite external sources.

Rules, filters and checks that keep an AI system within safe, intended limits. IBM describes them as policies, technical controls and monitoring that govern what models produce. 39 Site glossary

A fluent model output that presents false or fabricated information as if it were factual.

When AI states something false with confidence. NIST calls it confabulation; OpenAI calls hallucinations "plausible but false statements." 1 , 2 Site glossary

Under the EU AI Act and similar frameworks, an AI use that can seriously affect health, safety, or fundamental rights and therefore faces extra duties.

An AI system in an Annex III use case or a safety component of a regulated product, subject to quality management, data governance, logging, and human oversight duties.

A design pattern where a person must review or approve AI outputs before they take effect in a high-stakes workflow.

A design pattern where a person can intervene in an automated process but does not approve every step in advance.

A hyperscaler is a company that runs cloud services at very large scale. A hyperscale data center is one of its very large facilities. LBNL counts hyperscale as its own category of data center. 11 Site glossary

A form of liquid cooling in which servers sit submerged in a liquid that does not conduct electricity. LBNL lists immersion alongside direct-to-chip and rear-door cooling as liquid cooling options. 11 Site glossary

Procedures for detecting, containing, reporting, and remediating harmful AI behavior or misuse.

Using a trained model to produce an answer. Every time you send a prompt and get a reply, that is inference. Training is the earlier, much larger job of building the model. Site glossary

International standard (2023) for an AI management system that organizations may implement and optionally certify against.

A prompt crafted to make an AI ignore its safety rules. OWASP describes it as a form of prompt injection in which the input causes the model to disregard its safety protocols. 40 Site glossary

Units of energy used over time. A kilowatt-hour is 1,000 watts running for one hour, and a terawatt-hour is one billion kilowatt-hours. LBNL estimated U.S. data centers used 176 TWh in 2023, or 4.4% of U.S. electricity. 11 The IEA estimates data centers worldwide used about 415 TWh in 2024, about 1.5% of global electricity. 7 The two figures cover different areas, so they should not be compared directly. Site glossary

The date after which a model has no information from its training. Anthropic publishes a "reliable knowledge cutoff" for each Claude model. 3 Site glossary

The engine behind chatbots. A model trained on huge amounts of text to predict the next word, which lets it write and answer questions. 1 Site glossary

The delay between asking for something and getting a response, measured in time. Cloudflare notes that latency is a measure of time, not of how much data is moved. 41 Site glossary

Cooling computer chips with liquid, which carries heat away better than air. LBNL calls it an emerging technology for cooling dense equipment such as AI servers, with forms that include rear-door heat exchangers, direct-to-chip and immersion. 11 Site glossary

A neural language model trained on large text corpora to predict tokens and generate or transform language.

A way of building AI in which software learns patterns from examples instead of following hand-written rules. IBM calls it the subset of AI focused on algorithms that learn patterns from training data and then make predictions about new data. 42 Site glossary

Methods that fit models to data so systems improve at a task without being fully hand-coded for every case.

Software that translates text or speech from one language to another. Modern tools such as DeepL and Google Translate use AI models and handle whole documents, but they can still miss tone, idioms or legal meaning, so have important translations checked by a person. 43 Site glossary

Meta’s AI assistant. Meta says it is available on iOS and Android, on the web and desktop, on AI glasses and in its apps. 44 See Which AI? . Site glossary

The trained AI system itself. Companies give models names and version numbers (for example, a newer model may be faster or more capable). Site glossary

A short, structured disclosure of a model’s intended use, training data summary, evaluation results, and limitations (industry practice; formats vary).

Attacks that try to recover training data or approximate a proprietary model from its outputs.

An inventory of models in use, with owners, versions, and approved purposes.

AI that works with more than one kind of input or output, such as text, images, audio and video. Site glossary

AI built to do a specific job or set of jobs, such as translating or recognizing faces. IBM says the term describes nearly all current AI, in contrast with the hypothetical general AI. 16 Site glossary

Non-consensual intimate images. Federal TAKE IT DOWN Act and many state laws address NCII including certain AI-generated “digital forgeries.”

Intimate images shared or created without the depicted person’s consent, including some AI-generated forgeries.

Software made of many simple connected units, arranged in layers, that learns patterns from data. IBM describes it as a machine learning model that stacks “neurons” in layers and learns weights from data. 45 Site glossary

National Institute of Standards and Technology, a U.S. agency that publishes voluntary AI risk and measurement guidance.

A U.S. government guide that helps organizations manage AI risks. NIST says it is intended for voluntary use and was released on January 26, 2023. 46 Site glossary

The field of methods for analyzing and generating human language.

Non-binding intergovernmental principles on trustworthy AI first adopted by OECD members in 2019 and updated since.

Running a model on hardware controlled by the organization or user so data need not leave that environment.

Software or servers run on a company’s own equipment in its own buildings, not rented from a cloud provider. Site glossary

The Open Source Initiative defines it as an AI system released so that anyone can use it for any purpose, study how it works, modify it, and share it. It says open source models and weights must also come with the data information and code used to produce them. 47 Site glossary

Open-source usually means code under an OSI-style license; open-weight means parameters are released, which may use a different license.

A model whose trained settings (its weights) are published so that anyone can download and run it. That is not the same as open source: under the Open Source Initiative’s definition, the data information and code used to build the model must be shared as well. 47 Site glossary

Choosing to leave something you are included in by default. For AI, it often means turning off the setting that lets a company use your chats to improve its models. OpenAI, for example, says individual ChatGPT users can choose whether their conversations help improve its models. 23 Site glossary

White House Office of Science and Technology Policy.

The learned numbers inside a model that determine how it turns an input into an output. Training adjusts them. IBM describes them as the learned values within a machine learning model. 48 Site glossary

The learned numeric values inside a neural network that encode patterns from training.

An AI assistant built around answering questions with sources. Perplexity says its answers link to numbered citations. 49 See Which AI? . Site glossary

Adapting outputs using a user’s history or profile; raises distinct privacy and records questions in government use.

Details that can identify you, such as your name, address, phone number or account numbers. A good rule is to keep PII out of chatbots unless you know how the service handles it. Site glossary

Information that can identify a person, alone or combined with other data.

Inserting malicious or misleading data into training or retrieval corpora to alter later behavior.

Stages after pretraining (supervised fine-tuning, preference optimization, tool-use training) that shape how a model follows instructions and refuses requests.

A long-term contract to buy electricity from a specific power plant at an agreed price. EPA says renewable physical PPAs usually run 10 to 20 years, and that a financial (or “virtual”) PPA involves no physical delivery of power to the buyer. 50 Site glossary

A measure of how much extra electricity a data center uses beyond its computers, such as for cooling. LBNL defines it as the facility’s total electricity use divided by the electricity used by the IT equipment. 11 A PUE of 1.0 would mean no overhead; lower is better. Site glossary

The process of buying goods or services, including contract terms for AI tools used by government.

What you type or say to an AI. Clear, specific prompts get better answers. Site glossary

Writing prompts carefully, with clear context, instructions and examples, to get better answers. See Prompting . Site glossary

An attack that hides instructions in text an AI reads, such as a web page or document, so it follows them instead of yours. OWASP defines it as user prompts altering a model’s behavior in unintended ways, even when the input is invisible to humans. 40 NIST lists it among attacks on generative AI. 1 Site glossary

Information about origin, authorship, and transformations of content or data.

Under the EU AI Act, a person or organization that develops an AI system or general-purpose AI model and places it on the market or puts it into service under its name.

Laws that can make prompts, outputs, and related logs subject to disclosure and retention rules.

A data-center efficiency ratio of total facility power to IT equipment power; closer to 1.0 means less overhead.

Reducing the numeric precision of model weights to shrink size and cost, often with some accuracy trade-off.

A tall metal frame that holds servers stacked on top of each other in a data center. Data center capacity is often counted in racks. Site glossary

A method where an AI looks up information in a trusted source first, then writes its answer using what it found. AWS describes it as having the model reference an authoritative knowledge base outside its training data before it responds. 51 Site glossary

A model that works through a problem in steps before it answers. OpenAI says reasoning models use internal reasoning tokens before producing a response, which can improve results on harder tasks but uses more tokens. 52 Site glossary

Deliberately trying to make an AI misbehave, so the problems can be found and fixed before release. NIST describes AI red-teaming as exercises that look for potential adverse behavior or outcomes of a model and how they could occur. 1 Site glossary

A developer-stated plan that ties higher model capability to stronger safety measures before further scaling.

In some privacy and automated-decision laws, a person’s ability to obtain meaningful information about the logic involved in a decision that significantly affects them.

Grouping AI uses by potential harm so controls can scale with impact.

A training step in which people rate or rank an AI’s answers, and the model is adjusted toward the answers people prefer. OpenAI’s 2022 paper describes fine-tuning a model first on human demonstrations, then further with reinforcement learning from human feedback. 53 Site glossary

A structured argument, with evidence, that a system is acceptably safe for a defined use; referenced in some frontier-model policy discussions.

A supervised environment where organizations test AI systems with regulators under relaxed or clarified rules for a limited time.

A computer that stores data or runs programs for other computers over a network. A data center is, at its core, a building full of servers. Site glossary

Staff use of consumer AI tools without organizational approval or logging.

An AICPA attestation report on a service organization’s controls relevant to security, availability, processing integrity, confidentiality, and/or privacy under the Trust Services Criteria; commonly requested in vendor due diligence.

Evaluating people based on social behavior or personal traits to produce a score that leads to detrimental treatment; prohibited in specified forms under the EU AI Act and restricted in some U.S. state laws such as Texas TRAIGA.

The best published result on a given benchmark at a stated time.

A hypothetical AI that would be smarter than humans at nearly every task. It does not exist today: IBM calls it a hypothetical software-based system, and AWS calls it a theoretical concept. 8 , 9 Today’s chatbots are narrow AI, not super intelligence. The term is in the news because the White House now uses it: on Sept. 29, 2026 President Trump signed an executive order directing the federal government to use “Super Intelligence” instead of “artificial intelligence,” according to CBS News. 10 On this site, “AI” stays the name for the technology you can use today. See also the news item and the myth . Site glossary

Artificially generated data used for training, testing, or privacy-preserving analytics.

Audio, image, or video generated or substantially altered by AI. Policy issues include disclosure, provenance, deepfakes, and NCII.

Documentation covering an AI system (model plus tools, filters, and product policies), broader than a single model card.

Hidden instructions a company gives a chatbot before you type anything, telling it how to behave, what tone to use and what to avoid. Site glossary

A legal or platform process requiring removal of specified unlawful content within a deadline after valid notice.

A sampling setting that affects how deterministic or varied a model’s next-token choices are.

A chat mode that stays out of your history and isn't used for training. Names and rules differ by app; see Privacy . Site glossary

A structured description of who might misuse a system, how, and with what impact.

A chunk of text, often part of a word, that a language model reads and writes one at a time. 1 Site glossary

A chip designed by Google specifically for AI workloads. Google describes TPUs as custom-developed chips used to accelerate machine learning workloads. 54 Site glossary

Texas Responsible Artificial Intelligence Governance Act (HB 149), effective Jan 1, 2026.

The long, expensive process of building a model by showing it huge amounts of data and adjusting its parameters until it gets good at its task. It happens before you use the model; using it is called inference. Site glossary

Many AI chips wired together so they can work on training one model at the same time. The wiring matters because the chips must constantly share results. Site glossary

The last date of data included in a model’s training corpus, after which the base model has no built-in knowledge.

The text, images and other material a model learns from before you ever use it. Site glossary

A contractual or product setting stating that a customer’s prompts, files, or outputs will not be used to train or improve a vendor’s models.

The design behind most of today’s language models. It was introduced in a 2017 research paper, “Attention Is All You Need,” which described a network built solely on attention mechanisms. 55 Site glossary

Texas Risk and Authorization Management Program: cloud security authorization required before covered Texas state agencies buy certain cloud services.

Storage optimized for similarity search over embeddings, often used in RAG systems.

Using AI to copy someone's voice from a recording. The FTC warns a short clip can be enough. 6 Site glossary

USGS defines withdrawal as water removed from the ground or diverted from a surface source for use. Consumption is the part of that water that evaporates or is otherwise not available for immediate reuse locally. 56 A data center can withdraw much more water than it consumes, so check which one a number refers to. Site glossary

Hiding a marker inside AI-made images, audio, video or text so it can be detected later. Google says its SynthID tool embeds invisible watermarks and is designed to survive changes such as cropping or compression. 57 A missing watermark does not prove something is real. Site glossary

Signals embedded in or associated with model outputs to help detect synthetic origin; methods and robustness vary; often paired with provenance standards.

A measure of how much water a data center uses per unit of computing. LBNL defines it as the facility’s total water consumption divided by its IT equipment’s electricity use, reported in liters per kilowatt-hour. 11 Site glossary

When a chatbot looks things up online before answering and shows links to its sources. 4 , 5 Site glossary

Setting up a chain of steps that runs by itself when something happens, for example: a new email arrives, AI summarizes it, and the summary is posted to your team chat. Tools such as n8n and Zapier do this, and many now add AI steps. 58 See AI Systems . Site glossary

An arrangement some AI providers offer approved business customers in which the provider does not keep customer prompts and answers beyond processing them (details vary by provider). OpenAI says its Zero Data Retention excludes customer content from its abuse-monitoring logs and is subject to prior approval. 59 It is not something a free consumer account normally has. Site glossary

A vendor configuration stating that prompts and outputs are not stored for training or longer-term logs beyond short operational needs; confirm in the contract and DPA.

Asking a model to perform a task with no examples (zero-shot) or a small number of examples (few-shot) in the prompt.