AI-system security
Adversarial inputs and vulnerable integrations can undermine an AI application’s intended behavior.
Reviewed · 2026-10-072 cited sources
Skip to Observatory contentPSYBORG / OBSERVATORY V1Follow the evidence.
Follow the evidence.
See the connections.
Risks, incidents and research in context. An independent, curated map of AI’s human consequences.
A small historical collection, maintained editorially. Not exhaustive, not a live feed, and not a ranking of dangerous companies or models. Most recent record review: 2026-10-07.
01 / THE RISK LANDSCAPEQuestions worth
Open the registry ↗Questions worth
following.
Biometric misidentification
An incorrect identity match can become a harmful human decision.
Reviewed · 2026-10-072 cited sources
Privacy in AI deployments
Collecting and using personal data creates questions about purpose, access and retention.
Reviewed · 2026-10-072 cited sources
02 / DOCUMENTED EXPERIENCELook at
All incident records ↗Look at
what happened.
Confirmed occurrence and attributed allegations carry different labels. These historical examples do not measure incident frequency.
AI-generated voice in New Hampshire robocalls
The FCC documented a campaign using a cloned presidential voice and misleading caller identification.
Record date · 2024-011 cited source
FTC allegations about Rite Aid facial recognition
A regulator complaint describes harmful false matches in a retail surveillance deployment.
Record date · 2012 – 20201 cited source
New to the evidence library
Explore research ↗GPT-4 release documentation and limitations
A historical developer account of capabilities and limitations at launch.
Reviewed · 2026-10-071 cited source
NIST AI Risk Management Framework 1.0
Voluntary guidance connecting governance, context, measurement and management of AI risks.
Reviewed · 2026-10-071 cited source
NIST Generative AI Profile
A companion resource for applying AI risk-management guidance to generative systems.
Reviewed · 2026-10-071 cited source
Regulatory context
Explore regulations ↗EU Artificial Intelligence Act · original 2024 text
The original adopted text of Regulation (EU) 2024/1689, connected to its risk-related provisions.
Record date · 2024-06-131 cited source
Model documentation
Explore models ↗GPT-4 · March 2023
A historical model record tied to its original research announcement.
Record date · 2023-03-141 cited source