en.Wedoany.com Reported - A bipartisan group of U.S. House lawmakers recently introduced a bill proposing new regulations for the development of frontier artificial intelligence models, focusing on preventing cyberattacks and fraud, as well as promoting AI literacy workforce training. The 269-page draft of the "Great American AI Act" requires large frontier AI developers with annual revenues exceeding $500 million to publish model risk assessment frameworks and reports. It also formally codifies the "Center for AI Standards and Innovation" (CAISI) under the National Institute of Standards and Technology (NIST), which will be responsible for certifying "independent verification organizations" that audit AI companies' compliance with transparency requirements.
According to the bill summary, large frontier developers must hire licensed independent verification organizations to validate compliance with their frameworks and ensure the adequacy of frameworks and procedures. These independent verification organizations must obtain sufficient access to company materials and report audit results to the Center for AI Standards and Innovation. The bill is led by Representatives Jay Obernolte (R-California) and Lori Trahan (D-Massachusetts), and co-sponsored by Representatives Suhas Subramanyam (D-Virginia), Scott Franklin (R-Florida), Scott Peters (D-California), and Erin Houchin (R-Indiana).
This measure is seen as one of Congress's most ambitious attempts at AI regulation to date, but it has immediately sparked controversy. Civil society organizations, AI safety advocates, and labor groups criticize the bill for preemptively replacing state AI laws. One advocacy group argues that preventing states from "addressing emerging AI harms" would be a "generational mistake." On Capitol Hill, Democrats criticize the preemption provisions in the bill, while Republicans argue that regulation could stifle innovation.
In formally authorizing the Center for AI Standards and Innovation, the bill plans to provide it with a $300 million budget for fiscal years 2027 to 2029 and allow it to hire key technology experts at salaries exceeding traditional government pay caps. The bill also requires the center to collaborate with the Cybersecurity and Infrastructure Security Agency (CISA) on open-source security. CISA must provide funding to developers of critical U.S. open-source software packages for patch fixes, security assessments, and regular maintenance. AI companies must provide these developers with advanced models capable of detecting and fixing vulnerabilities, a provision that could significantly improve the cybersecurity of the open-source ecosystem, which has been plagued by a surge of AI-generated vulnerability reports in recent years.
Another provision requires NIST and the Department of Energy to collaborate with federal laboratories and the private sector to create AI safety testing platforms. These research centers will evaluate the capabilities and weaknesses of AI models and organize hackathons for public model testing. The bill also requires the Government Accountability Office to audit security measures protecting AI model weights and the security posture of the open-source ecosystem.
The legislation would also reauthorize the Cybersecurity Information Sharing Act, allowing companies to exchange threat intelligence with government agencies without facing legal liability or antitrust issues. Congress temporarily extended CISA's authorization until the end of September this year, but business groups and cybersecurity experts note that advances in AI are one of the most pressing reasons to place the program on a more permanent footing. Two major technology trade groups, the Information Technology Industry Council and the software industry association BSA, praised lawmakers for including the reauthorization of the Cybersecurity Information Sharing Act in the AI bill.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
