Data Created by AI Creates Legal Blackholes
Organizations needed to surround any data created by AI with a “tight matrix of contractual obligations” if they are to retain any commercial rights over its use, according to an Australian privacy lawyer.
Speaking at a Gartner conference in Sydney in March, special counsel Jim Lennon from law firm Norton Rose Fulbright said that because data created by AI had not been created by humans, then it raised copyright issues which were potentially “outside the legal containers” of data law.
Lennon used two examples to illustrate his point: the legal fight over a "selfie" image of a Macaque monkey taken by a wildlife photographer, and a legal judgment against Australian telco Telstra for the use of directory information.
In the case of the Macaque selfie, the image was taken by a camera owned by British photographer David Slater on a visit to Indonesia in 2011.
After taking numerous photographs, Slater left the camera within reach of the monkeys and one, who came to be known as Naruto, snapped his image.
Subsequently, animal rights organization PETA took Slater to court and won an initial judgment, which said that because the photograph had not been taken by a human, then Slater was not entitled to copyright.
The photographer appealed and subsequently reached an agreement with PETA to donate 25% of the proceeds of the image to a wildlife fund.
Lennon also used the example of Telstra, which uses machine techniques to assemble its White and Yellow Pages directories.
Back in 2010, the telco lost a legal case. The judgment established that Telstra had no claim on the copyright because machine techniques were used to create the directories.
“Data generated by AI, and insights generated by AI from data supplied to it may not be copyright,” Lennon said. "AI is proving intractable conceptually, and one of the difficulties in trying to establish a means of facilitation or control over the data is that you are not sure what the qualities of the data are which come out of AI systems."
This was likely to become particularly important, he added, as autonomous vehicles using AI become common on the roads.
While in terms of copyright ownership could be problematic, the responsibility could rest with a long chain of developers, manufacturers, and licensees who had created the vehicle.
"In this situation, we are faced with a situation where ownership and control are fraught with AI, and even more fraught are the contractual issues," Lennon said.
In these situations, the only approach was to “sew a tight matrix of contractual obligations” around the AI and the data it created and acted upon.
