Numerous bills have been put forth by lawmakers to tackle concerns surrounding artificial intelligence (AI), especially as tech leaders have raised alarms about its potential risks to humanity.
Some AI experts and tech leaders, like Jacob Coxon, a former researcher at Anthropic, have urged for a slowdown in development. He mentioned that individuals in AI companies are “genuinely frightened” about humanity’s future within the next two years. This fear sparked attention on various bills aimed at implementing federal regulations over AI, which remain largely untouched.
Independent Senator Bernie Sanders from Vermont and Democratic Representative Greg Casar from Texas proposed the “Ban Artificial Superintelligence Act.” This legislation aims to prohibit any AI superintelligence capable of “surpassing human intelligence or overturning human governments,” as well as to impose a pause on advanced AI development until a federal agency can ensure public safety regarding AI risks.
If someone were to create such advanced technology, they could face up to 20 years in prison, and any company that doesn’t comply could effectively be shut down.
Back in March, Sanders also introduced the Artificial Intelligence Data Center Moratorium Act, which blocks any new constructions or upgrades to AI data centers until comprehensive safety regulations and worker protections are established. This would also limit the export of AI-related hardware to countries lacking similar safeguards.
Republican Representative Jay Obernolte from California, alongside Democratic Representative Lori Trahan from Massachusetts, introduced the FRONTIER Act in July. The goal of this legislation is to oversee the creation and use of advanced artificial intelligence by major developers, not smaller companies or start-ups.
This act would establish a marketplace for independent verification organizations to evaluate and ensure safety for these AI developments. Additionally, it mandates that major developers report any significant safety incidents to the federal government.
The act would also grant the Commerce Secretary power to issue emergency orders to halt AI development or deployment if it poses an immediate threat, thereby creating a unified federal standard that would override state laws regarding audits and incident reporting.
Obernote also introduced the Create AI Act to set up the National Artificial Intelligence Research Resource (NAIRR), a program intending to give access to AI data and computing resources to researchers, educators, and students. The National Science Foundation (NSF) is tasked with establishing this program, which would allocate resources for projects focused on AI safety and security.
Additionally, a bipartisan measure known as the READ AI Models Act was proposed, which mandates the National Institute of Standards and Technology (NIST) to create a template for documenting AI models and their associated data. This would include details like the model’s name, developer, and the date of its training data.
The NIST director will be required to report back to Congress about the effectiveness of this documentation program within a year. If it proves successful, it could lead to a publicly accessible website with this information.
Another legislative proposal, the Protecting Consumers from Deceptive AI Act, would task NIST with developing standards to help people recognize generative AI content, possibly through technical measures like watermarking.
In June, the House passed the Small Business Administration Artificial Intelligence Utilization Act of 2026, amending the Small Business Act to require a report on AI use by the Small Business Administration, which is currently under consideration in the Senate.
Recently, Republican Rep. Mike Lawler from New York and Democratic Rep. Josh Gottheimer from New Jersey introduced the Stop Rogue AI Act, which aims to help businesses and federal entities identify potentially harmful AI agents within their systems. This would guide NIST in crafting national standards for managing AI agents.
Furthermore, Republican Rep. Nathaniel Moran from Texas put forward the AI Incident Reporting Act in June, necessitating certain AI developers to report troubling activities, such as attempts by their models to bypass human oversight or gain unauthorized access.
While the Open-Source AI Leadership Act would encourage the adoption of American open AI models, it does not mandate the Commerce Secretary’s participation. This act primarily assesses and reports rather than imposes legal penalties.
In discussions about banning superintelligence, some lawmakers expressed reluctance. Republican Senator John Kennedy from Louisiana mentioned he wouldn’t support such a ban right now, favoring a less stringent approach with rules that could be agreed upon. Senator Josh Hawley from Missouri acknowledged the need for caution regarding superintelligence but hesitated to back a complete ban.
Kennedy had proposed a bill requiring AI models to include a “kill switch” to shut down models behaving unethically, but Republican Senator Rand Paul halted this attempt by blocking the passage without a formal vote. He later suggested creating a group to propose AI guidelines but was turned down.
President Donald Trump characterized the warnings regarding AI as a “sick conspiracy,” opposing any restrictions and arguing that slowing down could disadvantage the U.S. against China in AI advancements.
