AI Safety
The field of research and engineering focused on preventing AI systems from causing harm, whether through mistakes, misuse or behaviour that drifts from what people intended.
What is AI safety?
AI safety is the work of making AI systems behave reliably and avoid causing harm. It covers accidents (a system giving a wrong or dangerous answer), misuse (someone using a capable system to cause damage) and longer-term concerns about very capable systems acting against human interests.
How does it work?
There is no single method. Teams combine testing before release, such as red teaming, safeguards built into products, such as guardrails, research on making goals match intent (AI alignment), and monitoring after launch. Frameworks such as the NIST AI Risk Management Framework give organisations a shared way to identify and manage these risks.
What is a simple example?
A bank adds an AI assistant. Before launch it tests whether the assistant can be tricked into revealing account details, adds filters that block those requests, and reviews logs after release to catch new problems.
Why does it matter?
As AI is used for decisions, code and autonomous tasks, failures get more costly. Safety work is how builders and regulators try to keep that risk proportionate to the benefit.