Generative AI art apps present many of the same ethical issues as AI writing apps because they operate in a similar way. They consume huge amounts of art and then create an image based on the user's prompt. This has enabled many of us who aren't artistically inclined to be able to create imagery that may be beyond our capabilities with pens or brushes, and has also led to new forms of computer-generated art, and all of that is great. However, as with ChatGPT, the creators of these art apps didn't ask artists or museums if they could consume their art, and thus they are ignoring intellectual property rights. Yes, you can use a generative AI art app to make art in the style of Georgia O'Keefe, Jean-Michel Basquiat, or Edward Hopper, but did they give you permission to do so?
Deepfakes present even greater ethical issues. Deepfakes are falsified videos or audio clips that are easy for AIs to produce and very difficult to detect. They've been used to make fake pornographic videos of celebrities, spread disinformation, smear innocent people as racists which led to death threats, and generate false political advertisements that threaten the validity of our elections (see case studies below). We are beginning to see steps to control deepfakes on both the national and corporate level: the UK has banned the creation & dissemination of non-consensual deepfakes, the European Parliament's AI legislation requires deepfakes to be watermarked, and Google's search engines will filter out explicit deepfakes from results.
Ref: Heikkilä, Melissa. "Google Is Finally Taking Action to Curb Non-Consensual Deepfakes." Technologyreview.com. MIT Technology Review, 6 August, 2024. https://www.technologyreview.com/2024/08/06/1095774/google-is-finally-taking-action-to-curb-non-consensual-deepfakes/
In early 2024, Eric Eiswert, a Baltimore-area high school principal received death threats and needed police protection after the release of a deepfake recording of him making racist and antisemitic remarks. The deepfake was apparently created by the school's athletic director as retaliation against Eiswert for an investigation into improper use of school funds. The athletic director is now facing criminal charges over the recording.
Thankfully, Eiswert was not harmed, but he had been placed on leave by his school system, his reputation had been stained by the deepfake, and he was forced to deal with the fear of death threats, all over an AI recording.
Ref: Looker, Rachel. "Baltimore High School Teacher Arrested Over Deepfake Racist Audio of Principal." BBC News. BBC.com, 26 April, 2024. https://www.bbc.com/news/world-us-canada-68907895
In January, 2024, Steven Kramer, a political consultant, paid another man a mere $150 to create an AI-generated robocall that sounded like President Joe Biden, telling New Hampshire voters not to participate in their state's Democratic primary. Kramer now faces more than two dozen criminal charges, along with a $6 million Federal Communications Commission charge, and the company that he used to transmit the call, Lingo Telecom, faces an additional $2 million fine.
Kramer claims he was trying to warn the electorate of the dangers represented by AI audio and video deepfakes. And while this may be a flimsy defense for his actions, his fake call does demonstrate exactly that: AI programs represent an unprecedented threat to our political process.
Ref: Ramer, Holly & Ali Swenson. "Political Consultant Behind AI-Generated Biden Robocalls Faces $6 Million Fine and Criminal Charges." APNews.com, Associated Press, 23 May, 2024, apnews.com/article/biden-robocalls-ai-new-hampshire-charges-fines-9e9cc63a71eb9c78b9bb0d1ec2aa6e9c
This video by AI expert Brent Anders introduces the TRAP Test for dealing with deepfakes:
Think Critically--question any digital image, audio or video
Realistic/Reliable/Reputable--Does this seem real? Is the source reliable & reputable?
Accurate/Authority--Are all parts accurate? Is the source an authority on the subject?
Purpose/Propaganda--What is the purpose of this? Is it likely to be propaganda?
