AI Safety
AI safety is the field concerned with preventing accidents, misuse and other harmful consequences of AI systems, from today's failures to catastrophic risks from future advanced AI.
What is AI Safety?
The International AI Safety Report, chaired by Yoshua Bengio and published in 2025, groups risks into malicious use, malfunctions and systemic risks. AI safety spans technical research, evaluation, governance and security.
How it works
It uses alignment research, testing, monitoring, safeguards and regulation.
← All Artificial Intelligence & Superintelligence definitions · ← Back to the A–Z definitions