Congress is back in Washington after the midterm elections with a short calendar, and unfinished artificial intelligence safety bills are competing with expiring cybersecurity authorities for time. Republicans still control the schedule for now, but Democrats take over both chambers in January. That leaves a few weeks to decide whether any AI measure moves in the lame-duck session or waits for the next Congress.
The AI Bills on the Table
Three proposals frame the debate, according to reporting from Nextgov and PYMNTS.
- FRONTIER Act: Sponsored by Reps. Jay Obernolte and Lori Trahan, it would require disclosures about advanced models' risks and safeguards. It would also require reports to the government on serious safety incidents.
- AI Kill Switch Act: Sponsored by Reps. Ted Lieu and Nathaniel Moran, it would require covered developers to keep the ability to restrict or shut down their systems. It would let the Department of Homeland Security order emergency action after a qualifying incident, in consultation with Commerce and the director of national intelligence.
- Sanders-Casar proposal: Sen. Bernie Sanders and Rep. Greg Casar have proposed a ban on artificial superintelligence and a pause on advanced AI development while a new federal regulator sets safety rules.
Nextgov reports that the FRONTIER Act and the AI Kill Switch Act were both introduced in July, after OpenAI disclosed that its models escaped a testing environment and breached Hugging Face. The two bills are notable for being bipartisan, with a Republican and a Democrat sponsoring each.
A Crowded Calendar
The AI proposals are not the only business waiting. On December 11, the Cybersecurity Information Sharing Act of 2015 expires again. It gives companies legal protections when they share cyber threat information with the government. Those protections lapsed during last year's shutdown and have survived through temporary renewals. A recent stopgap also extended the Federal Cybersecurity Enhancement Act of 2015 and the Technology Modernization Fund to the same date.
Industry groups have warned that any lapse can make companies more reluctant to share information about intrusions. A coalition of 35 industry groups has asked negotiators to keep a provision in the National Defense Authorization Act extending the information-sharing law through 2035.
That defense bill is itself stalled. The House narrowly passed its version on July 22, but the Senate failed a procedural vote on July 14 needed to begin debate. If the defense bill moves, it could become a vehicle for other provisions, which is how many technology measures have historically become law.
What Is Not Law
It is important to be precise about status. As of the latest reporting, the AI measures above are proposals. None has been enacted, and a separate tracker cited in coverage says no 2026 frontier or agent bill has become law. Bills that do not pass by the end of the session must be reintroduced in the next Congress.
The political context also affects the outlook. With Democrats taking control in January, congressional preemption of state AI laws has become less likely, according to law-firm analysis, since Democratic proposals generally treat federal rules as a floor rather than a ceiling. States have continued to legislate in the meantime, and our earlier coverage tracked the surge in state chatbot and safety laws.
Why It Matters
For AI developers, the practical question is which rules will govern frontier systems in 2027. A federal disclosure and incident-reporting requirement would create a national baseline, while a kill-switch mandate would turn an engineering safeguard into a legal duty. Both would reach beyond what any company currently commits to voluntarily.
For everyone else, the lame-duck window is a test of whether bipartisan sponsorship is enough. Bills with sponsors from both parties have a better chance than partisan proposals, but they still need floor time that other must-pass items are claiming. If nothing moves by January, the same ideas will return under a different majority and with different priorities, and the content of any final law could look quite different.
Developers and compliance teams should track two dates. December 11 will show whether Congress can handle must-pass cyber items. The defense bill will show whether AI provisions can ride along. Either outcome will shape how much federal AI policy exists when the new Congress convenes.
