Data products
AI Training Data Services
Custom data collection, annotation & evaluation
Frontier Alignment
CoT reasoning, SME RLHF, SFT & red teaming
Agentic AI
Agent trajectories, RL environments & tool-use logs
Multimodal and Speech AI training data
Speech corpora, VLM annotation & voice-agent evaluation
Physical AI
LiDAR annotation, robotics trajectories & sensor fusion
Model Integrity
Hallucination benchmarking, bias & compliance audits
Research Services
LLM-as-a-judge, synthetic data & context engineering
Coding Datasets
Repo-level code, PR pairs & held-out benchmarks
Why Appen
30 years of AI data expertise
SOC 2 & ISO 27001 certified
500+ global locales covered
Independent model evaluation
OTS datasets
Enterprise data
Perspectives
Content
Blog
Expert commentary on frontier AI development
Research
Research papers and publications from Appen
Case Studies
How leading AI labs work with Appen
Podcasts
Expert conversations on AI and data
Webinars & Events
Live and on-demand sessions with our experts
Featured
Human in the loop
Master RLVR at Scale
Improving LLM Safety Evaluations
About
Company
Annotation Platform
AI Data Platform (ADAP) — automation meets human oversight
About Us
30 years pioneering AI data
Investor Relations
ASX-listed: APX
News
Latest announcements and press
Careers
Build the future of AI with us
Get in touch
Join our workforce
Join our workforce
Get in touch
Home
Resources
Press External
When AI Models Hallucinate Not Just Text But What They See
Press
When AI Models Hallucinate Not Just Text But What They See
Sergio Bruccoleri
•
May 2026
Read article
The AI Innovator
