legal-document-to-plain-english-translator/”>Try Free Now: Legalese tool without registration

Find a LOCAL lawyer

## AI legalese decoder: Uncovering Potential legal Issues in AI Model Training

The methods used by tech companies to train the majority of generative AI models are likely to be ÔÇ£illegalÔÇØ, according to a senior executive who stepped down from British startup Stability.

Ed Newton-Rex, who recently resigned as VP of audio at London-based AI unicorn Stability over ethical concerns, told UKTN that the ÔÇ£jury is still outÔÇØ on whether companies like OpenAI and Google are breaking copyright law by training their models on creatorsÔÇÖ data.

Prior to his resignation, Newton-Rex was running Stability AIÔÇÖs audio department, which was working on a music generation tool. He left Stability because he did not agree with ÔÇ£the companyÔÇÖs opinion that training generative AI models on copyrighted works is ÔÇ£fair useÔÇÖÔÇØ.

It comes as rapid advances in generative AI are pushing the limits of copyright law. Intellectual property lawyers say most existing copyright laws were not written with AI in mind.

Newton-Rex said: ÔÇ£The juryÔÇÖs still out on whether this falls under the fair use exception to copyright law. But my view is that it doesnÔÇÖt.ÔÇØ

Referring to OpenAIÔÇÖs recent viral announcement of the video generation tool Sora, Newton-Rex highlighted what he sees as a common strategy by AI companies to avoid admitting theyÔÇÖre using copyrighted content.

ÔÇ£They said we trained on a lot of publicly available content,ÔÇØ Newton-Rex said. ÔÇ£Publicly available does not mean itÔÇÖs in the public domain ÔÇô publicly available means literally on the internet.ÔÇØ

He added: ÔÇ£You do not have the right to do anything you want with things on the internet, that is not how copyright works.ÔÇØ

Newton-Rex said this practice is ÔÇ£very widespreadÔÇØ and ÔÇ£a real problemÔÇØ.

Now running Fairly Trained ÔÇô a non-profit promoting ethical AI practices ÔÇô Newton-Rex told UKTN that ÔÇ£mostÔÇØ generative models rely on data sets from copyrighted work that ÔÇ£no one has consented toÔÇØ.

The legal precedent for AI models using content outside the public domain is still being determined. A handful of high-profile cases are set to cause ripple effects across the industry.

The most high-profile case is the New York Times lawsuit against OpenAI and Microsoft, which alleges that ChatGPT plagiarised the newspaperÔÇÖs content. Getty Images, meanwhile, is engaged in a lawsuit against Stability.

Stability has become one of the UKÔÇÖs most prominent AI companies, recently raising ┬ú89m in its Series A round.

The company has previously stated that ÔÇ£AI development is an acceptable, transformative and socially beneficial use of existing content that is protected by fair useÔÇØ.

Josh Little, a partner at the law firm Marriott Harrison told UKTN: ÔÇ£The debate over whether or not the training of large language models on unlicensed data is illegal does not come gift-wrapped with a straightforward answer. But itÔÇÖs also not a question thatÔÇÖs going away any time soon.ÔÇØ

### UKÔÇÖs AI approach is ÔÇÿna├»veÔÇÖ

AI technology crosses geopolitical borders, adding to the complexity of establishing how copyright law should be applied in the age of AI. Newton-Rex believes the UKÔÇÖs approach to AI policy is ÔÇ£incredibly lazyÔÇØ and ÔÇ£na├»veÔÇØ.

The UK government has repeatedly stated it does not plan to legislate AI in the ÔÇ£short termÔÇØ, instead hoping that a light-touch regulatory approach will help establish Britain as an AI powerhouse.

ÔÇ£ItÔÇÖs an incredibly lazy approach,ÔÇØ said Newton-Rex. ÔÇ£ItÔÇÖs a naive approach because it ignores the fact that you can have AI development, and you can have fair treatment of creatives.ÔÇØ

Newton-Rex said there doesnÔÇÖt need to be new legislation to protect artists, writers and other content creators.

ÔÇ£Just make it clear that training in this way is not permitted under the law,ÔÇØ he said. ÔÇ£You donÔÇÖt need a new law for that, you just point to an existing law and say no, thatÔÇÖs not permitted.ÔÇØ

By utilizing the AI legalese decoder tool, companies can identify and rectify potential legal issues in their AI model training processes. This tool can analyze the data sources used for training, ensuring compliance with copyright laws and ethical practices. By incorporating the AI legalese decoder into their AI development workflow, companies can mitigate risks associated with unauthorized data usage and uphold ethical standards in AI innovation.

legal-document-to-plain-english-translator/”>Try Free Now: Legalese tool without registration

Find a LOCAL lawyer

Reference link