Big Story: Startups need balanced copyright frameworks to compete in AI
As policymakers dig into whether and how to write rules around AI, much of the conversation is missing how proposals—including licensing requirements—would create barriers for startups to compete in the space. At a Senate Judiciary subcommittee hearing this week, lawmakers argued that AI companies should have to procure licenses before using copyrighted works to train large language models, a departure from the current legal landscape, where courts have held that fair use applies to AI training. In proposing licensing requirements, lawmakers are ignoring the reality that only larger companies would be able to comply, making it harder for startups to build models and compete in the AI space.
Under current interpretations of copyright law, AI developers are able to train their models on data—including data that may be protected by copyright. That’s crucial for startups training AI models, which require large and diverse training datasets to function effectively. Given the wide universe of content that is eligible for copyright protection—everything from in depth blog entries, to random Internet posts, to comments on a web forum—it would be impractical, if not impossible, for startups to identify every rightsholder and pay individual licenses, especially when any single work may only contribute a tiny fraction of a model’s training data. Recent court rulings have reinforced this interpretation, as a federal judge held that using copyrighted works to train AI models is protected under fair use, highlighting that models learn from patterns to generate transformative outputs. Upholding the principles of fair use protects the foundation that allows startups to innovate and compete.
Policymakers pushing licensing mandates as the answer misses the mark and could risk creating rules that make it harder for startups to build AI models. While Subcommittee Chair Sen. Josh Hawley (R-Mo.) warned against letting “a few mega corporations… make billions of dollars in profits, maybe trillions, and pay nobody for it,” the hearing overlooked how those same licensing proposals would make it nearly impossible for startups to build and scale with AI. For example, several media companies—including The Atlantic, Associated Press, Vox Media—reportedly entered into a licensing agreement with OpenAI. As policymakers consider legal frameworks around AI and training data, it’s crucial they keep in mind the limited budgets and legal compliance resources.
Policy Roundup:
House passes crypto legislation. The House passed three digital asset bills—the GENIUS Act, CLARITY Act, and the Anti-CBDC Surveillance State Act—which set new rules for web3 and crypto startups, including guidelines for when a digital asset is considered a security vs commodity and the agencies that have enforcement power. Also this week, the Senate Committee on Agriculture heard testimony from digital asset industry leaders who expressed their support for federal regulations overseen by the U.S. Commodity Futures Trading Commission which oversees commodities. Clearer digital asset regulation could lower compliance costs and give startups and investors more confidence to build within the U.S.
U.S. opens trade probe into Brazil, citing speech restrictions. The office of the U.S. Trade Representative launched a sweeping Section 301 investigation into Brazil this week, mentioning a wide array of grievances, from ethanol access to content takedown orders by Brazilian courts. While concerns over Brazil’s draconian intermediary frameworks are a legitimate target for U.S. trade policymakers, the volatile, inconsistent nature of this administration's trade moves risks productive, longstanding progress to resolve the barriers faced by U.S. Internet platforms.
Hearing explores responses to bad actors’ use of AI. A House Judiciary subcommittee held a hearing this week exploring how criminals are using AI and how AI can be used to combat crime. Witnesses and members underscored the importance of AI literacy within education, public-private collaboration, and using AI to defend against vulnerabilities as helpful steps policymakers should consider. And lawmakers were cautioned to avoid hamstringing U.S. innovation in response to AI concerns, because criminals don’t follow the rules anyways and many of them are outside of U.S. jurisdiction.
Massachusetts advances digital likeness bills. A Massachusetts committee moved forward with H.1615 and H.1751, which would create new property rights to a person’s digital likeness. Ahead of the Joint Judiciary subcommittee hearing on Tuesday, Engine submitted letters opposing the bills, which could install liability on AI developers for generating unauthorized digital replicas. As with other state-level digital replica proposals, these bills risk adding to a growing patchwork of laws that increase compliance burdens for startups.
On the Horizon:
TUE 7/22: The House Judiciary Committee will convene a hearing to discuss foreign abuse within U.S. courts, including third-party litigation funding at 10:00 AM ET.
TUE 7/22: The Senate Finance Committee will convene to vote on Joseph Barloon’s nomination to be Deputy U.S. Trade Representative and U.S. Ambassador to the World Trade Organization at 9:45 AM ET.
WED 7/23: The House Small Business subcommittee on oversight, investigations, and regulations will convene a hearing to discuss the Small Business Administration’s disaster assistance programs at 10:00 AM ET.
WED 7/23: The Senate Small Business & Entrepreneurship Committee will convene a hearing to discuss threats to innovation at 2:30 PM ET.
THU 7/24: The Federal Communications Commission will convene an open commission meeting to discuss removing barriers to broadband deployment and opening more spectrum at 10:30 AM ET.
Startup Roundup:
#StartupsEverywhere: Chicago, Illinois. Many businesses lack comprehensive cultural insights to inform their strategic decisions. Kyle Desana and his co-founder utilize advanced algorithms to analyze cultural signals on social media, allowing businesses to make informed decisions and predict key metrics. We sat down with Kyle to discuss his company, how AI is accelerating innovation, and the need to enhance capital access in startup ecosystems like Chicago.
