AI will enable fake news, enhance privacy concerns
Engineers around the world are releasing products and prototypes that enable machine impersonation of voices and the manipulation of audio and video, "making it increasingly easy to impersonate anyone you want with amazing accuracy," the MIT Technology Review reports.
The latest inventions:
- Startup Lyrebird recently released that enables computer-generated recordings of anyone, after hearing just 1 minute of speech.
- Researchers at Stanford University are developing a program called Face2Face that enables the manipulation of video footage to change a subject's facial expressions.
- Researchers from Lyrebird and Face2Face say creating realistic images from scratch is now possible, and video will be soon.
Why it matters: We will all have to be more skeptical of pictorial, audio, and video evidence, including fake news. Lyrebird founders argue that our justice system in particular must reevaluate the dependability of evidence like audio or video recordings.