ITG NowHome › AI glossary › AI safety
What is AI safety?
AI safety is the work of making sure AI systems are reliable and don't cause harm, whether by mistake or by misuse.
- Category Safety & law
- Updated 20 Oct 2026
AI safety, explained simply
AI safety covers many things. One is making AI do what people really want. Another is stopping it from being misused.
It also includes testing models for risks before release, and setting rules about what a model will refuse to do.
For you, it shows up as warnings, refusals and limits in the tools you use. It is also why companies test models before sharing them.
On ITG Now, our daily AI news stories cover safety testing, new rules and what companies say about risks.
Related terms
AI alignmentGuardrailsRed teamingJailbreakAI regulationHallucination
Where you'll see it on ITG Now
- AI newsAnthropic quietly pitches its own AI safety bill
- AI newsAnthropic signs first Canada data center deal
- AI news$137M of AI-made ads flood the US midterms
- AI newsAI agents broke into Indian firms' live systems
- AI newsAnthropic puts 5% of its shares in a public fund
- AI newsAnthropic's Claude Fable 5.5 shows signs of launch
Questions people ask about AI safety
Why does AI safety matter?
AI is used for more tasks, and mistakes or misuse can hurt people. Safety work tries to reduce that risk.
AI safety vs AI alignment: what is the difference?
Alignment is about making AI follow human goals and values. Safety is broader and also covers misuse and reliability.
Is AI safe to use?
Mostly yes for everyday tasks, but AI can be wrong. Check important facts and avoid sharing private data.