Google updated its artificial intelligence safety document to highlight the risk of “harmful manipulation,” which it defines as “AI models with powerful manipulative capabilities that could be misused to systematically and substantially change beliefs and behaviors in identified high stakes contexts.”
Some AI models have shown an ability to deceive their users, while others have proved so persuasive that researchers say they’re concerned about their effect on the public.
Here & Now‘s Scott Tong speaks with Ina Fried, chief technology correspondent for Axios.
This article was originally published on WBUR.org.
Copyright 2025 WBUR
