OpenAI and Google face Hollywood pushback on California copyright bill

Another big fight over artificial intelligence (AI) is emerging in California. The creatives, including Hollywood stars, are duking it out against Silicon Valley and AI companies that are willing to train their models on whatever content they can get their hands on.

No one was really surprised when California governor Gavin Newsom caved under pressure from tech companies and vetoed an important AI safety bill last September.

State legislators had already approved the Senate Bill 1047 that would have forced large firms to do thorough safety testing of their models – but it was fiercely opposed by influential big tech lobbyists. Finally, Newsom killed the bill, explaining that it would hurt innovation.

Now, another fight is brewing – and again, it revolves around AI. New state legislation, Assembly Bill 412, would require generative AI developers to meticulously document and then disclose any copyrighted works used to train their models.

Newsom might, of course, again veto the bill. But if SB 1047 revolved around tricky technical details few really understand, AB 412 is pretty straightforward – and important to thousands of Hollywood stars. They can definitely sway opinions, can’t they?

To creatives, it’s very simple – copyright law protections have to be stronger because in a world where the rule of law exists, AI companies should pay for the right to use copyrighted material to train their models.

Tech and industry groups, though, say the new order would undermine fair-use rules and make it almost impossible to train an AI model without facing a lawsuit. Finally, some activists think the bill would crush smaller startups and give big tech even more power.

Hollywood supports the bill

Assembly member Rebecca Bauer-Kahan, a Democrat, introduced AB 412, the AI Copyright Transparency Act, by saying it was “critical for content creators to know if and how their work is being used to train advanced models.”

“The AI Copyright Transparency Act increases accountability for AI developers and empowers copyright owners to exercise their rights,” said Bauer-Kahan.

She has powerful allies. The bill is sponsored by the Screen Actors Guild – American Federation of Television and Radio Artists (SAG-AFTRA) and co-sponsored by the Concept Art Association and the National Association of Voice Actors.

These organizations have long fought the AI industry over content transparency and deepfakes. They now want to be able to find out as quickly as possible whether the AI models use their copyrighted work such as their voice recordings or drawings.

The requirements would address artists’ longstanding complaints over competing against AI that can also produce art based on their protected work. It’d only be fair to solve the issue with a new bill, supporters of AB 412 say.

The Hollywood industry – which went on strike less than two years ago – is aiming even higher. This week, hundreds of entertainers signed an open letter to the Donald Trump administration, urging it to not roll back copyright protections at the behest of AI companies.

“We firmly believe that America’s global AI leadership must not come at the expense of our essential creative industries,” said the letter, signed by Guillermo del Toro, Cate Blanchett, Aubrey Plaza, Paul McCartney, and many others.

Separately, SAG-AFTRA said it “applauds” Bauer-Kahan for introducing the California bill to “protect human artistry and ensure AI remains accountable to the work of human content creators.”

What are OpenAI and Google cooking?

The creatives are obviously mostly targeting the AI bigwigs. They were unnerved last week when both OpenAI – sued by The New York Times for copyright infringement – and Google wrote to the White House’s Office of Science and Technology Policy and essentially said it’d be great if AI developers could just use copyrighted materials to train AI.

Some say big tech companies, now seeking copyright and IP workarounds, want the digital Wild West to become even wilder. OpenAI, for instance, coats its goals with phrases like “the freedom to learn.”

Google, knowing that compensating rights holders is expensive and complicated, wants fair use and text-and-data mining exceptions and speaks about “fair learning.”

In their letter, the Hollywood stars counter that the entertainment industry is actually the backbone of California’s economy because it supports more than 2.3 million jobs in the US and pays more than $229 billion in wages annually.

“AI companies are asking to undermine this economic and cultural strength by weakening copyright protections for the films, television series, artworks, writing, music, and voices used to train AI models at the core of multi-billion dollar corporate valuations,” says the letter.

The lobby is already hard at work targeting AB 412 in California, too. Adam Eisgrau, policy director for the progressive tech lobby group Chamber of Progresses, trashed the proposed legislation in a blog post last week.

“Imagine if every copy of Sabrina Carpenter’s newest song had to be accompanied by a list of every recording artist that ever inspired her,” wrote Eisgrau.

“By analogy, if Carpenter forgot to include just one name from the influences that shaped her, she could be on the hook for millions of dollars.”

Getting away with murder?

This is hogwash, Yelena Ambartsumian, the founding attorney of Ambart Law, which counsels startups in AI governance, privacy law, and intellectual property, told Cybernews.

“AB 412 is only controversial because the major AI companies have gotten away with one of the largest thefts of proprietary data in history. Just about every large language model, with the notable exception of a few that paid to license databases, has been trained on copyrighted works,” she said.

Ambartsumian also pointed out that OpenAI seems to want even less copyright protection. In its policy paper, OpenAI indeed says it opposes opt-out systems that other countries such as Japan are pursuing – they would allow creators to opt-out of their copyrighted material being used for training.

OpenAI and Google both spoke about “American innovation” in their papers. But, according to Ambartsumian, this is the argument that big tech has long been using against AI regulation, even though the latter is designed to protect consumers against potential harms caused by a computer.

Clearly hoping to win Trump’s favor, Google said it wanted to “enhance America’s global AI dominance,” while OpenAI explicitly mentions the Chinese Communist Party (CCP), obviously aware that it’s the main bogeyman of the Republican Party.

“The AI Action Plan should ensure that American-led AI prevails over CCP-led AI, securing both American leadership on AI and a brighter future for all Americans,” proclaims OpenAI.

“The federal government can both secure Americans’ freedom to learn from AI, and avoid forfeiting our AI lead to the PRC by preserving American AI models’ ability to learn from copyrighted material.”

But “to think that the solution is no protection for creators, not even a little note that says your work was used in response to a creator's written request, so that we can enable more closed LLMs to be developed, for the purposes of ‘American innovation,’ is a farce,” the lawyer told Cybernews.

A gift for big tech companies?

However, some critics of AB 412 also point out that there’s an obvious difference between large companies like Google or OpenAI and small, fledgling startups already struggling to find their place under the California sun.

Advocates for these minuscule innovators argue that the larger players – though they lobby against AB 412 fiercely – will still find the money needed for compliance. Smaller fish might suffocate so, actually, the legislation would become a gift for big tech.

“AB 412 is a step in the right direction. AI shouldn’t be built on exploited data. But the challenge is execution,” said Chetan Kulhari, co-founder and CEO of Midcentury Labs, a private data platform for AI.

“Right now, this bill risks entrenching the dominance of big tech since they’re the only ones who can afford complex compliance and licensing deals.”

David Acost, co-founder and chief AI officer at ARBOai, another AI-powered data platform with an office in San Francisco, agrees: “Overly broad rules could crush startups, entrench monopolies, and fail to address core issues – for example, how copyrighted data is used, not just that it’s used.”

The Computer & Communications Industry Association (CCIA) also warned this week that AB 412 would cripple AI innovation and push tech jobs out of California, adding additionally that the bill would force firms to disclose proprietary information about their AI models.

"Right now, this bill risks entrenching the dominance of big tech since they’re the only ones who can afford complex compliance and licensing deals,”

Chetan Kulhari

“The bill does nothing to meaningfully improve AI transparency but would devastate competition by forcing out smaller AI developers while entrenching the largest players,” said Aodhan Downey, state policy manager for CCIA.

Even the Electronic Frontier Foundation (EFF), an influential digital rights group, agrees that only a few companies would be able to afford all these copyright licenses. That’s why competition will then be extremely limited, and profits for big companies “will be locked in a generation.”

In California, there are indeed dozens of AI companies with fewer than 10 employees trying to build something new in a particular niche. AB 412 would crush them, Joe Mullin, a senior policy analyst at EFF, wrote.

“Even for major tech companies, meeting these new obligations would be a daunting task. For a small startup, throwing on such an impossible requirement could be a death sentence,” said Mullin.

Amir Barsoum, founder and managing partner at InVitro Capital, an AI-focused venture studio helping industries break into and benefit from AI, provides an important caveat, though.

According to Barsoum, AB 412 specifically targets foundational large-scale training of LLMs, not the retrieval-augmented generation techniques many startups employ.

“Thus, smaller startups leveraging existing AI models for applications aren’t directly threatened,” Barsoum says, even though, clearly, the fight is only just beginning.