Amazon in the US has been accused by an ex-employee of breaching its own internal copyright rules as it sought to keep up with rivals in the race to develop ever more sophisticated AI models. When she challenged the policy breaches, she claims, she was told it was fine because “everyone else is doing it”.
The claim is made in a discrimination and unlawful dismissal lawsuit filed by AI researcher Dr Viviane Ghaderi. Although it’s a small part of the litigation, the allegation seems to confirm concerns in the music and wider creative industries that those developing AI models are being very slack when it comes to their copyright obligations.
The core part of Ghaderi’s lawsuit deals with the discrimination she claims she was subjected to when she returned to Amazon after maternity leave. That discrimination first resulted in her being demoted and then in her dismissal.
On her return to the company she also “inherited a project relating to defendants’ large language models”, the lawsuit explains. “Ms Ghaderi was in charge of flagging violations of Amazon’s internal copyright policies and escalating these concerns to Amazon’s in-house legal department”.
In March 2023, the legal filing says, Ghaderi met with her team director, Andrey Styskin, who wanted “to understand why defendants were not meeting goals on a project relating to search quality on the Alexa team. Ms Ghaderi outlined the challenges she had faced because of Amazon’s internal copyright-related policies - which she had fully complied with”.
Styskin allegedly “rejected Ms Ghaderi’s concerns about Amazon’s internal policies and instructed her to ignore those policies in pursuit of better results because ‘everyone else’ - ie other AI companies - ‘is doing it’”.
Ghaderi says she communicated her concerns to Amazon’s legal team that managers were directing her to violate both the company’s own policies and intellectual property law.
She also notes that, since she raised those issues internally at Amazon, various AI companies have been sued by copyright owners. She specifically cites the New York Times’ lawsuit against OpenAI, though she could have referenced the lawsuit filed by a group of music publishers against the Amazon-backed Anthropic. These lawsuits, she says, prove the concerns she raised were reasonable.
Most of the AI companies that have been sued by copyright owners argue that they don’t need permission when using copyright protected works to train an AI model because such use is ‘fair use’ under US law.
Needless to say, the copyright owners do not agree. If they prevail in court, claims like those made by Ghaderi will likely be used to argue that many of those involved in the big AI race were rather reckless when it came to navigating their legal obligations.