AI copyright lawsuit: ETtech Explainer: Nvidia hit with AI copyright lawsuit

Three authors have filed a lawsuit against graphics processing unit (GPU) giant Nvidia for illegally training its artificial intelligence (AI) platform NeMo on their copyrighted works. This is another development in the growing tensions globally involving intellectual property right holders and companies that make AI models, resulting in similar lawsuits that saw entities such as Microsoft and OpenAI cross swords.

Here’s an explainer of what the Nvidia lawsuit says and what these concerns are.

Who are the authors?

Authors Brian Keene, Abdi Nazemian, and Stewart O’Nan proposed a class action lawsuit against Nvidia in a San Francisco federal court on Friday.

Keene’s 2008 novel ‘Ghost Walk’, Nazemian’s 2019 novel ‘Like a Love Story’, and O’Nan’s 2007 novella ‘Last Night at the Lobster’ are the works allegedly used as data input for Nvidia’s AI platform NeMo, Reuters reported.

Nvidia describes NeMo as a ‘toolkit’ for conversational AI and natural language processing. Users can leverage the ‘building blocks’ the platform provides to efficiently create their own generative AI.

Gen AI is built using large language models which are trained on vast amounts of data. Major AI players like OpenAI have said that gen AI models do not infringe copyright as they are “transformative” in nature.ETtech Explainer: Legal tussle between OpenAI and NYT

What are the allegations?

The authors argue that their books were included in a dataset of about 196,640 books used to train NeMo to simulate ordinary written language.

They were taken down in October for ‘reported copyright infringement’. The authors say that this proves Nvidia admitted to the alleged violation.

The lawsuit seeks damages from Nvidia, of an unspecified amount, for people in the US whose works were used without permission to train NeMo in the last three years, Reuters reported.

With the surge in demand for AI, the GPU-maker’s stocks have seen an unprecedented rally in recent weeks. It became the third-largest company in the world behind Microsoft and Apple, with its market capitalisation crossing $2.3 trillion.

Also read | Nvidia on cusp of overtaking Apple as second-most-valuable company

Growing concerns

Previously, ChatGPT-maker OpenAI and its major investor Microsoft have been sued by the New York Times and several Pulitzer-winning authors for allegedly training their AI models on copyrighted material worth billions of dollars.

In India, ET reported on January 26 that news publishers are seeking changes to the Information Technology Rules for protection against copyright infringement in the process of training AI models and ensuring fair compensation.

Some companies have entered licensing deals to use data legitimately. For instance, Google in February cut a $60 million per year deal with community forum website Reddit for real-time access to its data for AI training.

Three authors have filed a lawsuit against graphics processing unit (GPU) giant Nvidia for illegally training its artificial intelligence (AI) platform NeMo on their copyrighted works. This is another development in the growing tensions globally involving intellectual property right holders and companies that make AI models, resulting in similar lawsuits that saw entities such as Microsoft and OpenAI cross swords.

Here’s an explainer of what the Nvidia lawsuit says and what these concerns are.

Who are the authors?

Authors Brian Keene, Abdi Nazemian, and Stewart O’Nan proposed a class action lawsuit against Nvidia in a San Francisco federal court on Friday.

Keene’s 2008 novel ‘Ghost Walk’, Nazemian’s 2019 novel ‘Like a Love Story’, and O’Nan’s 2007 novella ‘Last Night at the Lobster’ are the works allegedly used as data input for Nvidia’s AI platform NeMo, Reuters reported.

Nvidia describes NeMo as a ‘toolkit’ for conversational AI and natural language processing. Users can leverage the ‘building blocks’ the platform provides to efficiently create their own generative AI.

Gen AI is built using large language models which are trained on vast amounts of data. Major AI players like OpenAI have said that gen AI models do not infringe copyright as they are “transformative” in nature.ETtech Explainer: Legal tussle between OpenAI and NYT

What are the allegations?

The authors argue that their books were included in a dataset of about 196,640 books used to train NeMo to simulate ordinary written language.

They were taken down in October for ‘reported copyright infringement’. The authors say that this proves Nvidia admitted to the alleged violation.

The lawsuit seeks damages from Nvidia, of an unspecified amount, for people in the US whose works were used without permission to train NeMo in the last three years, Reuters reported.

With the surge in demand for AI, the GPU-maker’s stocks have seen an unprecedented rally in recent weeks. It became the third-largest company in the world behind Microsoft and Apple, with its market capitalisation crossing $2.3 trillion.

Also read | Nvidia on cusp of overtaking Apple as second-most-valuable company

Growing concerns

Previously, ChatGPT-maker OpenAI and its major investor Microsoft have been sued by the New York Times and several Pulitzer-winning authors for allegedly training their AI models on copyrighted material worth billions of dollars.

In India, ET reported on January 26 that news publishers are seeking changes to the Information Technology Rules for protection against copyright infringement in the process of training AI models and ensuring fair compensation.

Some companies have entered licensing deals to use data legitimately. For instance, Google in February cut a $60 million per year deal with community forum website Reddit for real-time access to its data for AI training.