pre-deployment testing
In plain wordsChecking a new AI model for dangerous abilities and bad behavior before it is released to the public, sometimes with outside experts or government testers given early access.
The evaluation of an AI model or system before release to identify dangerous capabilities, misaligned behavior, security weaknesses and policy violations, and to decide whether extra safeguards, a staged deploymentstaged deployment, n.: Releasing a new AI system step by step — first to testers, then to limited groups, then widely — so that problems can be spotted and fixed before everyone has it. or a delay is needed. It typically combines automated benchmarksbenchmark, n.: A fixed, shared test — a set of questions or tasks with an agreed way of scoring — used to compare AI systems with one another and to track progress over time., dangerous capability evaluationsdangerous capability evaluation, n.: A test of whether an AI system can do things that could cause severe harm in the wrong hands or on its own — such as helping make a bioweapon, hacking computers or copying itself., internal and external red teamingred teaming, n.: Deliberately attacking your own AI system — trying to make it misbehave, leak secrets or help with harm — so that weaknesses are found before real adversaries or users find them., alignment auditsalignment audit, n.: A systematic investigation of an AI model to find out whether it has hidden goals or bad tendencies — such as deceiving its users — that ordinary testing might miss. and tests of mitigations such as classifiers, with results summarized in a system cardsystem card, n.: A public report released with an AI model that explains what it can do, how it was tested for safety, what risks were found and what protections were put in place.. Since 2023 governments have sought early access to frontier models: the UK and US AI safety institutes tested several models before release from 2024, and the EU's General-Purpose AI Code of PracticeGeneral-Purpose AI Code of Practice, prop. n.: A voluntary rulebook, published by the EU in 2025, that makers of large AI models can sign to show they follow the EU AI Act's rules on transparency, copyright and safety. (2025) calls for systemic-risk assessment, including model evaluations, before market placement. Critics note that pre-release windows may be too short for thorough testing, and that risks also arise after release, from fine-tuning, new tools and internal deployment.
The institute had nine days of pre-deployment testing, which was not enough for the agentic evaluations.
Origin
Pre-release safety testing became visible with OpenAI's staged release of GPT-2 (2019) and external red-teaming of GPT-4 (2022–2023). At the AI Safety Summit at Bletchley Park in November 2023, leading developers agreed that governments should be able to test next-generation models before deployment; the UK and US institutes published joint pre-deployment evaluations from late 2024.
Usage note
Government testers were renamed in 2025: the UK AI Safety Institute became the AI Security Institute, and the US institute became the Center for AI Standards and InnovationCenter for AI Standards and Innovation, prop. n.: The US government's AI testing and standards center, part of the National Institute of Standards and Technology. It evaluates advanced AI models, American and foreign, for national-security risks and works with companies on voluntary testing.. Testing of internally deployed models, used for research before any public release, is a growing concern.
Related
- dangerous capability evaluationdangerous capability evaluation, n.: A test of whether an AI system can do things that could cause severe harm in the wrong hands or on its own — such as helping make a bioweapon, hacking computers or copying itself.
- red teamingred teaming, n.: Deliberately attacking your own AI system — trying to make it misbehave, leak secrets or help with harm — so that weaknesses are found before real adversaries or users find them.
- staged deploymentstaged deployment, n.: Releasing a new AI system step by step — first to testers, then to limited groups, then widely — so that problems can be spotted and fixed before everyone has it.
- system cardsystem card, n.: A public report released with an AI model that explains what it can do, how it was tested for safety, what risks were found and what protections were put in place.
- AI Safety InstituteAI Safety Institute, n.: A government agency that tests the most advanced AI models for dangerous abilities and studies how to make them safe. Britain and the United States set up the first ones in 2023.
- third-party AI evaluationthird-party AI evaluation, n.: Testing of an AI system by someone other than the company that made it — a government institute, a nonprofit lab or independent researchers — to check what it can do and how risky it is.
- safety casesafety case, n.: A written, evidence-backed argument that a particular AI system is safe enough for a particular use, putting the burden of proof on the developer, as with nuclear plants or aircraft.
Mentioned in
red teamingred teaming, n.: Deliberately attacking your own AI system — trying to make it misbehave, leak secrets or help with harm — so that weaknesses are found before real adversaries or users find them., Center for AI Standards and InnovationCenter for AI Standards and Innovation, prop. n.: The US government's AI testing and standards center, part of the National Institute of Standards and Technology. It evaluates advanced AI models, American and foreign, for national-security risks and works with companies on voluntary testing., METRMETR, prop. n.: A nonprofit research group that checks new AI systems for dangerous skills, best known for tracking how long the tasks are that AI agents can now finish on their own., staged deploymentstaged deployment, n.: Releasing a new AI system step by step — first to testers, then to limited groups, then widely — so that problems can be spotted and fixed before everyone has it., system cardsystem card, n.: A public report released with an AI model that explains what it can do, how it was tested for safety, what risks were found and what protections were put in place., third-party AI evaluationthird-party AI evaluation, n.: Testing of an AI system by someone other than the company that made it — a government institute, a nonprofit lab or independent researchers — to check what it can do and how risky it is., Apollo ResearchApollo Research, prop. n.: An AI safety organization, founded in London in 2023, that tests advanced AI systems for signs that they might secretly pursue goals their developers did not intend., CybenchCybench, prop. n.: A test of AI agents' hacking skills built from 40 professional 'capture the flag' security puzzles, used to judge how much an AI could help cyberattacks or cyber defense.