A contentious provision of the 2025 budget reconciliation bill (dubbed the “Big Beautiful Bill” by its Republican supporters) that would have effectively banned state regulation of AI for 10 years has been removed after a bipartisan group of senators expressed concern that it would undermine federalism and paralyze state legislatures for too long in quickly evolving field.
The ban was initially included in the House Energy and Commerce Committee’s markup as a measure to bolster growth, as part of the GOP’s spending package seeks to slash taxes and curb government spending.
The original provision threatened to withhold $500m in Broadband Equity, Access, and Deployment (BEAD) funding for states that attempted to regulate AI within a ten-year period.
But concerns existed that the provision would violate the Byrd Rule, which states that reconciliation bills focus on budgetary concerns rather than policy.
Attorneys general from 40 states expressed concern that a blanket ban on AI state regulation would cause serious harm.
Eventually, the Senate voted 99-1 to remove the provision entirely, defeating a proposition by House Republicans to lower the ban to five years or allow carveouts for certain areas of AI regulation. The lone no vote was provided by Senator Thom Tillis (R-NC).
Attorneys general from 40 states expressed concern that a blanket ban on AI state regulation would cause serious harm, especially in a field that is so rapidly shifting. Some criticized the apparent decision to disable states regulatory prerogatives while coherent federal legislation still seemed distant.
But the nearly unanimous vote underscores the need for comprehensive AI regulation at the federal level to avoid a patchwork of potentially conflicting state regulations.
Reflecting the consensus opinion of the industry, OpenAI CEO Sam Altman welcomed the initial 10-year moratorium in absence of federal AI legislation, stating that “it is very difficult to imagine us figuring out how to comply with 50 different sets of regulation.”
Senator Marsha Blackburn (R-TN), who spearheaded the move to kill the ban, expressed frustration that there has been so little movement on the to enact federal AI legislation, and applauded states for taking up the slack, especially in critical areas like the protection of children.
NASAA, a group of state securities regulators, also hailed the decision to remove the freeze.
“NASAA applauds the Senate for sending a clear message that the states should not be forced to make an impossible choice between enforcing laws that foster innovation and fight fraud and accepting federal funding to expand broadband access,” it stated in a press release.
Meta’s new division to search for AI holy grail
Mark Zuckerberg, the CEO of Meta, plans to launch a new AI business unit called Meta Superintelligence Labs.
The division reportedly seeks to attain Artificial General Intelligence (AGI), which is traditionally defined as reasoning conceptually identical to human thought formation, yet with extremely sophisticated processing capacity.
The company’s new AI tack will overshadow its open-source Llama AI model, whose most recent iteration was met with a lackluster reception.
It remains uncertain whether true AGI can be achieved with foreseeable technology, or if pouring massive funding into research at the present stage will lead to a solution. Some prominent researchers believe AGI is achievable within a decade; others see no current viable path.
But overestimations of the nearness of AGI have been a perennial sticking point in the field, with predictions of imminent AGI development dating back to the 1970s.
Meta has already poached key executives from OpenAI, Anthropic, and Google’s DeepMind to staff the division. It will be co-headed by Alexandr Wang, CEO of AI data management startup Scale – an outfit in which Meta had recently invested $14.3 billion for a 49% stake – and former GitHub CEO Nat Friedman.
OpenAI previously claimed that Meta had offered bonuses as high as $100m to lure its employees to Superintelligence Labs. Zuckerberg has indicated that his company will spend at least $65 billion this year on AI development.
CMS plans to use AI to expedite Medicare approvals and curtail abuse
The Department of Health and Human Services’s (HHS) Centers for Medicare and Medicaid Services (CMS) has launched a pilot program that seeks to use AI to expedite certain prior authorization processes for Medicare, while also tackling waste, fraud and abuse in the system.
The Wasteful and Inappropriate Service Reduction (WISeR) Model will be overseen by The Center for Medicare and Medicaid Innovation (CMMI), and will partner with technology companies to focus on certain areas that are susceptible to unnecessary and low-value services, such as skin and tissue substitutes, electrical nerve stimulator implants, and knee arthroscopy for knee osteoarthritis, among others.
The Medicare Payment Advisory Commission notes that $5.8 billion in Medicare spending was spent on services that provided “minimal benefit.”
The CMS noted that while AI will be used to streamline the approval process, the ultimate approval decision will be made by a licensed clinician. The WISeR Model will launch on January 1, 2026, and will run for six years.
