Google, Openai, Andreessen Horowitz and a group of 400 entertainment industry celebrities were among those who submitted 8,755 comments on requests for ideas for the US AI Action Plan.
The plan’s comment period ends on Saturday (March 15th), and according to a White House statement, the comments will help “develop AI action plans to maintain and strengthen America’s global AI control.”
Among the comments were more than 400 Hollywood actors, directors, musicians, writers, and other insiders who signed an open letter to the Trump administration urging the government not to undermine copyright laws for AI model training.
Celebrities such as Cate Blanchett, Mark Ruffalo and Paul McCartney have denounced recommendations by Openai and Google for using copyrighted content in AI training to “remove all legal protections and existing guardrails.”
Such removals threaten 2.3 million American jobs and more than $229 billion annual wages in the arts and entertainment industry, they said.
“There is no reason to undermine or eliminate copyright protections that helped America flourish. It is not a time when AI companies can use copyrighted material simply by doing what the law requires.
This letter was sent to the National Science Foundation. This was seeking comment on the US AI Action Plan. According to the administration, the action plan will “define priority policy actions.”
The proposal elicited 8,755 comments, including letters from the entertainment industry.
Read more: Openai Calls Federal Government to Preempt State Laws
Openai, Google’s stance on copyright
Openai’s proposal includes having the AI model train copyrighted materials while protecting the rights of content creators.
According to the startup, “America’s robust and balanced intellectual property system has long been the key and key to global leadership in innovation.” “The federal government can ensure the freedom of Americans to learn from AI and avoid confiscating AI leads to the PRC (China).
Google’s proposal includes “balanced copyright rules” that allow for the use of copyrighted content in regards to “fair use and mining text and dates.”
“These exceptions allow us to use copyrighted, published material for AI training. We can avoid very unpredictable, disproportionate and long negotiations with data holders during model development or scientific experiments without having a major impact on the right sholder.”
VC company Andreessen Horowitz joined the discussion with its own submitted comments. It urged the government to clarify the “appropriate” relationship between copyright law and model development.
“Access to training data is currently the most important factor in determining global AI leadership,” the VC company wrote. However, the US “Legacy Media Companies are fighting wars, among other things, using data to train AI, and are erroneously pushing forward with this infringing copyright.”
Andreessen Horowitz argued that the creators’ work needs to be protected. However, the “active” interpretation of copyright law “puts US developers at a major disadvantage” against other countries with loose IP laws such as China, the company said.
The company recommended that the government make it clear that training of AI models on copyright work is permitted under current copyright laws. This is to avoid “unnecessary” lawsuits to stop “necessary” model training.
The VC company also filed a statement of interest while pending AI copyright lawsuits and stated that it should be described as “fair use” as model training is “transformative and does not use more data than necessary to train the model, thus placing no risk to the market for the original work.”
