"Responsible AI is not keeping pace with AI capability, with safety benchmarks lagging and incidents rising sharply. Documented AI incidents reached 362 in 2025, up from 233 in 2024."
"AI models have become quite good at programming, with scores on the SWE-bench test rising from 60 percent to close to 100 percent in a year. However, all AI models tend to be deficient in different areas."
"When attorneys use AI models to make over two dozen fake citations and misrepresentations of fact, it exemplifies the gap between AI usage and responsible practices."
Artificial intelligence has rapidly reached 53% adoption among the population within three years, surpassing previous technologies like personal computers and the internet. However, this surge has led to a rise in harmful AI incidents, with documented cases increasing from 233 in 2024 to 362 in 2025. The Stanford AI Index Report indicates that responsible AI practices are not keeping pace with advancements, as safety benchmarks lag behind. Many organizations and students report using AI, yet the technology still struggles with accuracy and reliability in various applications.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]