Featured Stories
Rand: When Disaster Strikes, Could AI Help? Q&A With Jessica Jensen
SANTA MONICA, California, Sept. 3 (TNSbrep) -- Rand issued the following Q&A on Sept. 2, 2026, involving senior policy researcher Jessica Jensen:
* * *
When Disaster Strikes, Could AI Help? Q&A with Jessica Jensen
When disaster strikes, the public sees the most visible parts of the response: evacuations, first responders, rescue operations, and recovery efforts. But behind the scenes is a much larger network of people working to help communities manage through chaos.
That job is getting harder. Disasters are more frequent, more severe, and more costly. Pressure is mounting on emergency managers
... Show Full Article
SANTA MONICA, California, Sept. 3 (TNSbrep) -- Rand issued the following Q&A on Sept. 2, 2026, involving senior policy researcher Jessica Jensen:
* * *
When Disaster Strikes, Could AI Help? Q&A with Jessica Jensen
When disaster strikes, the public sees the most visible parts of the response: evacuations, first responders, rescue operations, and recovery efforts. But behind the scenes is a much larger network of people working to help communities manage through chaos.
That job is getting harder. Disasters are more frequent, more severe, and more costly. Pressure is mounting on emergency managersand the organizations that support them.
In a new RAND study supported by the Markle Foundation as part of the AI for Disasters and Emergencies Initiative (AIDE), senior policy researcher Jessica Jensen and her RAND colleagues examined how artificial intelligence could help carry some of that burden.
Over more than two decades, Jensen has helped emergency management rethink how communities prepare for, respond to, and recover from increasingly complex disasters. Now, she is exploring how AI could help a field that's already stretched thin adapt to even greater demands.
Q: What could AI realistically help emergency managers do--and how could that improve outcomes for communities?
A: AI can help with prediction and modeling, including when events are likely to occur and how severe impacts may be. It can also help bring together data from different groups, so people better understand what is happening during a disaster. If AI is used well, it can help communities get the right help at the right time without inefficiency or duplicated effort, ultimately saving lives.
Before disasters happen, emergency managers spend a lot of time on routine work like preparing plans, creating outreach materials, and managing grants. AI can speed up much of that work. That allows emergency managers to spend less time behind their desks and more time doing important public-facing work--helping people prepare, building community relationships, and focusing on the things that lessen disaster impacts.
Your team identified more than 1,000 AI-enabled products relevant to emergency management. What do these tools do, and what surprised you about your findings?
We were shocked by the number of tools out there. We thought we might find dozens. Instead, we found 1,179.
A lot of them support response. They support situational awareness, hazard monitoring, damage assessment, information sharing, and more. But the market is uneven, with more tools for response than for preparedness or recovery.
A big challenge facing emergency managers is that many of the products are very narrow. For example, many offer one piece of insight, but they do not connect different information streams or help create a fuller picture. That creates what we call a stacking problem. People would need to use multiple tools and would have to connect them. The products also tend not to facilitate interorganizational coordination, a hallmark of emergency management task areas.
What are the biggest barriers to using these tools--and what will it take for emergency management to adopt them more widely?
As we found, it's not that the tools do not exist. It is whether organizations have the capacity to evaluate them, buy them, integrate them, and keep using them over time.
A lot of these products require technical expertise. Often, you need someone with an IT background to set the tool up and provide ongoing support. Many jurisdictions don't have that kind of staffing. Many products also require internet connectivity, which can become an issue during response. A significant number depend on another product to work. And pricing is often hard to find--which makes procurement difficult.
Then you add privacy or cyber concerns, and the reality is that many offices are already stretched thin. One issue by itself may be manageable, but taken together, they can be overwhelming.
This is also where there is room for progress. Technology companies can design products more specifically for this field and prioritize transparency on pricing, access, and technical requirements--which will drive better adoption. Policymakers can make decisions about governance, procurement, and funding that make adoption easier. And emergency managers can build more familiarity with what these tools can do and where their limits are.
In your research, what role did large language models (LLMs) and AI chatbots play in emergency management today?
We found they could be especially useful for administrative and communication tasks in emergency management. They can help summarize reports, draft materials, translate information, and process large amounts of text more efficiently.
Those kinds of general-purpose tools stood out as one place the field may be able to realize benefits quickly, since they have broad capabilities, are user-friendly, and already available to some jurisdictions. For more complex emergency management tasks, though, there is still more research and real-world testing that needs to be done to determine LLMs effectiveness.
This study was supported by the Markle Foundation's AI for Disasters + Emergencies (AIDE) Initiative. How critical is philanthropic support in a fast-moving area like this?
One of the most important things philanthropies can do is bring together technology companies, the user community, and the researchers developing the underlying technologies, and facilitate conversations about how to balance everyone's interests. That convening role--and the ability to sustain those conversations--can make the difference between tools that sound promising and tools that actually help communities. That is exactly what the AIDE Initiative is doing--bringing the right people together to explore how AI can better support disaster preparedness, response, and recovery.
You've dedicated your career to emergency management. What's kept you in the field so long?
I've known I wanted to serve this field since I was 14. Around that time, a regular at the restaurant where I worked would talk about his job helping communities and organizations recover after disasters. What stayed with me was how many different people and systems have to come together to help a community heal.
Now, years later, this work feels even more meaningful. There is an incredible opportunity in emergency management to keep evolving the practice to better serve communities. And I care deeply about the people who do this work. They are hardworking, passionate, and committed--even when resources are scarce and the demands keep growing. In a world of increasing disasters and rising impacts, that work matters enormously. It's what keeps me going.
* * *
Original text here: https://www.rand.org/pubs/commentary/2026/09/when-disaster-strikes-could-ai-help-qa-with-jessica.html
[Category: ThinkTank]
Center of the American Experiment Issues Commentary: Yet Another Minnesota Fraud? Natural Gas Innovation Act
MINNETONKA, Minnesota, Sept. 3 -- The Center of the American Experiment, a civic and educational organization that says it creates and advocates policies, issued the following commentary on Sept. 1, 2026, by policy fellow Darren Nelson:
* * *
Yet another Minnesota fraud? Natural Gas Innovation Act
Enhanced with AI
On the evening of August 26, the Center of the American Experiment (CAE) won a hard-earned and well-deserved award from the State Policy Network (SPN) for bringing to national attention Minnesota's fraud scandal of "Feeding Our Future." In his acceptance speech, CAE President John
... Show Full Article
MINNETONKA, Minnesota, Sept. 3 -- The Center of the American Experiment, a civic and educational organization that says it creates and advocates policies, issued the following commentary on Sept. 1, 2026, by policy fellow Darren Nelson:
* * *
Yet another Minnesota fraud? Natural Gas Innovation Act
Enhanced with AI
On the evening of August 26, the Center of the American Experiment (CAE) won a hard-earned and well-deserved award from the State Policy Network (SPN) for bringing to national attention Minnesota's fraud scandal of "Feeding Our Future." In his acceptance speech, CAE President JohnHinderaker pointed out that:
"It can be tough being a policy organization in a blue state because it isn't easy to get good policy implemented. On the other hand, bad policy and incompetent government create a lot of opportunities for important public campaigns."
One such "bad policy," from an "incompetent government" in this "blue state" of Minnesota, has "created" yet another "opportunity for an important public campaign" on fraud. That being the Natural Gas Innovation Act or NGIA, which is not about "natural gas innovation," but about natural gas elimination from residential homes and commercial businesses.
Although probably not actionable in a court of law, this fraudulently titled statute is actionable in the court of public opinion. And, the NGIA may not be within the letter, but is certainly within the spirit, of the legal definition of fraud:
"For a statement to be an intentional misrepresentation, the person who made it must either have known the statement was false or been reckless as to its truth."
Situation
Minnesota is now into the fifth year of a long-term state plan to effectively ban the use of natural gas for most homes and businesses. This plan is the NGIA, which was signed into law by Governor Tim Walz in June 2021.
The NGIA has an opaque end game: "It is the goal of the state of Minnesota that through the Natural Gas Innovation Act, utilities reduce the overall amount of natural gas produced from conventional geologic sources delivered to customers."
The NGIA has speculative means to that end: "Innovative resource means biogas, renewable natural gas, power-to-hydrogen, power-to-ammonia, carbon capture, strategic electrification, district energy, and energy efficiency."
The Minnesota Public Utilities Commission (PUC) is far clearer about NGIA than the statute:
"The NGIA and its innovation plans enable gas utilities to begin testing methods to reduce their emissions, and in some cases, transfer their business away from natural gas entirely. The results of approved pilots will lay the groundwork for future decarbonization efforts in the state and help gas utilities achieve the state's goal of economy-wide carbon neutrality by 2050. "
Six of the eight "innovative resources" use a lot of electricity. That six includes the most favored one of strategic electrification, which in turn includes geothermal heat pumps, which can result in four times the usage of electricity in winter.
And electricity in Minnesota is no longer cheap, since the era of renewables. From 2002 to 2025, electricity prices increased 118%, whilst natural gas decreased by 9%. Thus, electricity went from 319% more expensive than natural gas in 2002, to 763% by 2025. Renewable electricity caused this. The No-more Natural Gas Act will make this far worse.
Complication
The first complication is that the NGIA is one of the purportedly more than 40 state climate laws and programs in the Climate Action Framework (CAF), which aim to:
* reduce GHG emissions by 50% by 2030;
* 100% carbon-free electricity by 2040;
* achieve net-zero emissions by 2050.
The second complication is that the PUC is pursuing reduced natural gas use through other means. This includes the Energy Conservation and Optimization (ECO) program, which encourages heat pumps over central air conditioners for some new homes. This also includes the Gas Integrated Resource Plan (IRP), which will evaluate specific dual-fuel heating models that use electricity for most heating and switch to natural gas only during the coldest hours.
The third complication is that natural gas has increasingly been used for electricity generation in the 21st century, as renewables generation was massively increased, coal generation was retired, and no new nuclear generation has been allowed. The latter is more than likely to change in 2027 or 2028, but it might take until 2035 for new nuclear generation capacity to come into operation.
These complications make repealing the NGIA somewhat harder and possibly mooted. This is because these will still put pressure on residential and commercial users to reduce and replace natural gas use in favor of industrial, very large (e.g. data centers) and electricity generation users. The latter two will reinforce and accelerate that, given the long lead times for building new nuclear electricity generation.
Solution
These complications can be overcome. Firstly, affordability is the top voter issue in this state, and energy is one of the key reasons for that. Repealing the NGIA is an easy first step for the new governor and legislature to take in 2027. This is made easier given that Democrats across the nation are backing away from climate change as an election issue in 2026.
Secondly, NGIA's repeal is also made more likely by the current Minnesota government study to possibly end the moratorium on new nuclear power in the state. This study has strong bipartisan support and is due by the end of January 2027. Expectations are that the study will recommend ending the moratorium. And new nuclear could come online by 2030, rather than 2035, given the recent advances in technology, along with federal government support and data center uptake.
Repealing the moratorium and NGIA together could possibly form part of a legislative package on energy affordability. This would not only immediately defund NGIA related programs to curtail and replace natural gas, and related rate increases, it would send a strong signal to the PUC to back off and get real on ECO and IRP.
This might also sow the seeds for the PUC, CenterPoint and Xcel to heed the winds of change in the air on climate change. Big Tech have been a big part of that change, as they increasingly look to build nuclear and natural gas to power data centers, and to do so in ways that don't hurt ratepayers (e.g. bring their own power off-grid) or even help (e.g. build excess capacity on-grid).
Conclusion
We at CAE will:
* continue to expose the NGIA, for the fraud it is on Minnesotan voters and ratepayers;
* encourage public push-back on the NGIA, on CenterPoint, Xcel and the PUC;
* seek support for the repeal of the NGIA, amongst both DFL and GOP candidates.
Regarding the second, please consider having your say to the PUC about the NGIA by:
* 4:30pm October 14, in terms of CenterPoint's annual report to the PUC on their natural gas replacement plans, spending and related NG rate increases; and/or
* 4:30pm October 21, in terms of Xcel's annual report to the PUC on their natural gas replacement plans, spending and related NG rate increases.
Anyone from the public can and should comment. Not just on the details of the CenterPoint and Xcel reports, but also on the need to repeal the NGIA. One can submit comments here:
* Online: https://mn.gov/puc/get-involved/public-comments/, and follow the instructions;
* Email: consumer.puc@state.mn.us;
* U.S. Mail: Consumer Affairs Office, Minnesota Public Utilities Commission, 121 7th Place East, Suite 350, St. Paul MN 55101.
* * *
Original text here: https://www.americanexperiment.org/yet-another-minnesota-fraud-natural-gas-innovation-act/
[Category: ThinkTank]
Center of the American Experiment Issues Commentary: Minnesota Taxpayers Paid $3 Million for a 'Medicare for All' Cost Estimate - Where is It?
MINNETONKA, Minnesota, Sept. 3 -- The Center of the American Experiment, a civic and educational organization that says it creates and advocates policies, issued the following commentary on Sept. 1, 2026, by policy fellow Matt Dean:
* * *
Minnesota taxpayers paid $3 million for a 'Medicare for ALL' cost estimate. Where is it?
P.J. O'Rourke put the problem in one sentence in 1993: "If you think health care is expensive now, wait until you see what it costs when it's free." Proponents such as Abdul El-Sayed, Peggy Flanagan, and the Democratic Socialists of America still sell Medicare for All as
... Show Full Article
MINNETONKA, Minnesota, Sept. 3 -- The Center of the American Experiment, a civic and educational organization that says it creates and advocates policies, issued the following commentary on Sept. 1, 2026, by policy fellow Matt Dean:
* * *
Minnesota taxpayers paid $3 million for a 'Medicare for ALL' cost estimate. Where is it?
P.J. O'Rourke put the problem in one sentence in 1993: "If you think health care is expensive now, wait until you see what it costs when it's free." Proponents such as Abdul El-Sayed, Peggy Flanagan, and the Democratic Socialists of America still sell Medicare for All asa simple, cheaper, "free-at-the-point-of-service" replacement for private insurance. Unlike other states, Minnesota paid for a cost estimate for instituting Medicare for All and promised to deliver it to us on January 15, 2026.
But what does Medicare for All really mean? For Flanagan, "Peggycare" means eliminating all private health insurance and replacing it with a single-payer, taxpayer-funded program where you would all be essentially on whatever benefit set the government wants to allow you. For the past two decades, progressive advocates have moved past "If you like your plan, you can keep your plan" to "Medicare for all" (often abbreviated by the left as "M4A"). Just as "safe legal and rare" became "abortion is health care" the left's giant leap left in health care is broadening its reach and making the bet that more moderate voters are no longer necessary to secure wins. In places like Minnesota and Michigan and New York, the bet has paid off in recent years.
Under M4A-Peggycare arrangement all health care costs would likely be paid by some taxpayers, but not others. Progressives like Zorhan Mamdani propose shaking the money out of the pockets of overturned billionaires to pay the extra "small amount of tax increases." Nobody is quite sure how much, but the billionaires will pay for it, according to the socialists.
The Congressional Budget Office estimates the cost of PeggyCare to cost up to an additional $3 trillion/year nationally. But this is of course back of the napkin math. If only we had a real study that looked at the cost of socializing health care in Minnesota.
Luckily, the taxpayers of Minnesota hired a legislature who saw fit to spend $2.975 million on a study to answer that question and let us know on January15, 2026.
In 2023, DFL majorities could not pass the Minnesota Health Plan -- the single-payer bill introduced as HF 2798 / SF 2740. So they turned it into a study and parked it in the health and human services conference report. Laws of Minnesota 2023, chapter 70, article 16, section 19 ordered the Department of Health to contract for a 10-year cost-benefit analysis of that exact proposal versus the current system. The report was due January 15, 2026.
The original study language carried $1.2 million of your money to fund the study. The enacted appropriation was $1.815 million in FY 2024 and $580,000 in FY 2025, with a $580,000 base in FY 2026 -- nearly $3 million authorized,-- to answer a simple question: what would this plan cost, and who would pay.
January 15 came and went. MDH told the Legislative Reference Library the report would slip to "later in the spring of 2026." Spring ended. Summer ended. As of September 1, 2026, the Library still lists the mandate as received: nothing. No public model. No accounting of coverage, underinsurance, system capacity, total spending, jobs, or disparities -- the very items the statute required.
* * *
Original text here: https://www.americanexperiment.org/minnesota-taxpayers-paid-3-million-for-a-medicare-for-all-cost-estimate-where-is-it/
[Category: ThinkTank]
Center of the American Experiment Issues Commentary: Back to School, Back to the Union?
MINNETONKA, Minnesota, Sept. 3 -- The Center of the American Experiment, a civic and educational organization that says it creates and advocates policies, issued the following commentary on Sept. 1, 2026, by policy fellow Catrin Wigfall:
* * *
Back to school, back to the union?
It's that time of year again -- back to school! Teachers are diligently working on establishing classroom routines and writing lesson plans, but there is another important decision that falls during the busiest time of year: determining if union membership is right for them.
From today (Sept. 1) until Sept. 30, Minnesota
... Show Full Article
MINNETONKA, Minnesota, Sept. 3 -- The Center of the American Experiment, a civic and educational organization that says it creates and advocates policies, issued the following commentary on Sept. 1, 2026, by policy fellow Catrin Wigfall:
* * *
Back to school, back to the union?
It's that time of year again -- back to school! Teachers are diligently working on establishing classroom routines and writing lesson plans, but there is another important decision that falls during the busiest time of year: determining if union membership is right for them.
From today (Sept. 1) until Sept. 30, Minnesotateachers can exercise their right to opt-out of union membership and stop financially supporting Education Minnesota and its affiliates.
Union membership is a personal decision, and for a variety of reasons, thousands of Minnesota educators have said "no thanks" to what the union and its affiliates prioritize. A common theme is concern that union leadership has become increasingly disconnected from the professional realities teachers face every day. Educators have pointed to too much attention -- and funding -- going toward broader political causes that don't align with teachers' priorities and values.
The perceived gap between leadership priorities and member needs has raised questions over the value of membership. That's particularly true as dues continue to rise while liability insurance can be obtained elsewhere for a fraction of the cost and without the politics attached.
Saying "no thanks" to union membership doesn't make an educator less committed to his or her students or colleagues. It means that teacher is letting the union know it doesn't speak for every educator and shouldn't hold a monopoly on their voice.
* * *
Fast Facts for Educators
* Salary and benefits are provided by your employer, the school district, and are not impacted by union membership status.
* Your pension is not impacted by union membership status.
* Tenure and seniority are not impacted by union membership status.
* The union is not the only one that offers liability insurance and legal consult.
* The district and union cannot discriminate or retaliate against you based on your union membership status. You also cannot be fired from your job for not belonging to a union.
* No meeting with your local union rep is required to resign union membership.
* Non-members are not "free loaders" or "scabs" -- the union fought for and won the right to be the exclusive representative for all employees within the collective bargaining unit. If the union only wants to represent dues-paying members, it could lobby (as it does on many other issues) to have the law changed.
* If you opted out of the union in a previous year, and did not sign any new membership forms, your resignation is still in effect and you do not need to send in future opt-out letters.
Hear from some Minnesota teachers below who decided union membership is not right for them.
* * *
Original text here: https://www.americanexperiment.org/back-to-school-back-to-the-union-3/
[Category: ThinkTank]
American Action Forum Issues Insight: Rethinking the Open-source AI Debate
WASHINGTON, Sept. 3 -- The American Action Forum issued the following insight on Sept. 2, 2026, by technology and innovation policy analyst Angela Luna:
* * *
Rethinking the Open-source AI Debate
Executive Summary:
* Open artificial intelligence (AI) models - those that can be downloaded, inspected, and modified - offer greater access and flexibility of use, while lowering costs and barriers to AI innovation compared with closed models; many policymakers, however, favor closed models over open models based on the assumption that they offer greater safety and control.
* As open models - particularly
... Show Full Article
WASHINGTON, Sept. 3 -- The American Action Forum issued the following insight on Sept. 2, 2026, by technology and innovation policy analyst Angela Luna:
* * *
Rethinking the Open-source AI Debate
Executive Summary:
* Open artificial intelligence (AI) models - those that can be downloaded, inspected, and modified - offer greater access and flexibility of use, while lowering costs and barriers to AI innovation compared with closed models; many policymakers, however, favor closed models over open models based on the assumption that they offer greater safety and control.
* As open models - particularlythose from Chinese developers - become more capable and widely adopted by U.S. businesses and other organizations, their growing use is raising questions about the U.S.' AI leadership and national security tensions.
* Rather than treating open models as the problem, policymakers should take an evidence-based approach to managing the risks of increasingly capable models, while recognizing the value of both open and closed models in maintaining a diverse and competitive AI stack.
-
Introduction
Artificial intelligence (AI) models can be released along a spectrum, from closed (or proprietary) models controlled by their developers to fully open-source models whose components are available for download. In between are open-weight models, which make the model weights available but not necessarily other components. Both open-source and open-weight models can be referred to generally as "open" models. While open models offer greater access, flexibility of use and lower costs and barriers to innovation compared with closed models, many policymakers remain concerned that greater openness could make AI systems harder to secure and control.
As open models - particularly those from Chinese developers - become more capable, U.S. businesses, developers, researchers, and government organizations are increasingly using them to build and innovate with AI. Yet this growing adoption is raising new questions about U.S. AI leadership, as greater reliance on Chinese models could affect U.S. competitiveness, as well as national security, given potential risks related to data privacy, government use, and sensitive applications.
Rather than treating open models as the problem, policymakers should take an evidence-based approach to managing the risks of increasingly capable models, while recognizing the value of both open and closed models in maintaining a diverse and competitive AI stack.
Open-source AI and the Regulatory Dilemma
Open source has long been a driver of innovation. The early open-source software movement of the 1980s supported the internet and underlying systems by lowering the cost of software and allowing open sharing of knowledge. Today, AI developers and users are seeking to bring the same benefits to AI by developing open-source foundational models. As previous American Action Forum research reported, AI models can be released along a spectrum from closed to open source.
On one end are closed source, or proprietary, models, which are accessed through consumer applications or Application Programming Interfaces (APIs) that allow businesses to use models hosted on the developer's infrastructure. On the other end are fully open-source models, whose components are available for download, modification, and deployment on users' own infrastructure. Somewhere in between are "open-weight" models, which make the billions of parameters learned during training available for download, while other components, such as training code and data, usually remain unavailable to users. Both open-source and open-weight models can be referred to generally as "open" models. Table 1 shows the model release characteristics, with each approach offering different levels of access, data privacy, cost, customization, and control.
* * *
Table 1: The AI Model Release Spectrum
* * *
Because AI models, open or closed, contain powerful capabilities, each end of the spectrum offers its own benefits and risks. Supporters of proprietary models argue that they offer greater safety and control over deployment, while supporters of open-source models argue that openness provides greater access, lowers barriers to entry and costs for businesses, and enables the experimentation needed to foster AI innovation. This, however, is not a new debate among policymakers who have often favored proprietary models over open models based on the assumption that they offer greater safety.
The Growing Open-model Market
Open models have expanded significantly across the AI market. Stanford's AI index report shows that the number of AI-related projects on GitHub - a platform used by developers to share and collaborate on open-source software - grew from 1,549 in 2011 to approximately 5.6 million in 2025, and had a year-over-year growth of 24 percent from 2024 to 2025. Additionally, a report from OpenRouter shows that while proprietary models still dominate usage, open source reached approximately 33 percent of usage by late 2025.
Additionally, because companies find that open AI models are cheaper, these models are becoming an important part of the AI market. For example, a recent study found that proprietary models cost, on average, six times more than open models. The study also estimates that shifting demand from proprietary to open models could reduce average prices by more than 70 percent, potentially generating $24.8 billion in consumer and businesses savings in 2025.
Open models are also becoming more popular internationally. While U.S. companies such as Meta and Google have been major contributors to open model development, companies from China, the United Kingdom and France are gaining popularity. Notably, China's open-AI models are gaining momentum. Hugging Face (a platform of open-source AI systems) reports that among the most popular open models created in 2025, the majority were either developed in China or built from Chinese models. Alibaba's Qwen family, for example, has more than 113,000 derivative models - new models built or modified from Qwen - on Hugging Face. That's more than Meta and Google combined. Chinese open-weight models are also among the most-used models on OpenRouter, with five Chinese models ranking among the top models by weekly usage.
This trend has accelerated following the release of Chinese open models such as DeepSeek and Kimi K3, as their rapidly improving capabilities have both narrowed the performance gap with U.S. closed models and increased their popularity in the market. A report shows (https://a16z.com/asserting-american-leadership-in-open-source-ai/) that 80 percent of developers building with open-source tools are using Chinese tools, and this past January, the Alibaba Qwen family of models became the most widely adopted open AI systems in the world.
New Tensions: AI Leadership and National Security
As the capabilities and adoption of Chinese open-AI models continue to grow, the tradeoff is expanding to broader tensions of AI leadership and national security.
Open models - particularly those developed in China - are becoming increasingly capable, leading more developers and businesses to use them to build and innovate with AI. For example, telecom companies such as AT&T are using open models, including some developed in China, to replace proprietary models for specific workloads (which now power about 25 percent of the company's overall AI usage and expect it to reach 80 percent over time). AT&T reports this adoption gives them lower costs, greater control over their data, and more flexibility and choice among AI providers. Additionally, the recent Hugging Face and OpenAI incident - in which an agent powered by OpenAI models escaped its testing environment, gained access to the internet, and compromised parts of Hugging Face's infrastructure - exemplifies the cybersecurity utility these models give to U.S. businesses. Hugging Face reported that while it initially used a proprietary model, the model blocked the analysis because it triggered safety guardrails. The company later deployed an advanced open-weight model developed by Chinese firm Z.ai, allowing it to conduct the analysis while keeping sensitive data within its infrastructure.
But the same adoption that makes these models valuable also brings tradeoffs. The use of Chinese open models by U.S. companies raises some important questions, including concerns on data privacy, U.S. government and businesses' dependence on Chinese models, and the sensitive applications that these models could enable. Moreover, it raises questions over which models become the foundation of the global AI market. Because open models are increasingly driving AI innovation, the companies and countries that supply them could have greater influence not only in how the technology develops, but also in the standards that shape global AI adoption.
Restricting access to Chinese open models could also bring tradeoffs. Limiting access could reduce the range of models available to U.S. developers, businesses, and academia, hindering their ability to study, develop and integrate AI into their work, while also reducing competition by limiting alternatives to models offered by dominant proprietary developers in the United States. The conversation should therefore not be framed as a choice between closed or open models, but on how to build and strengthen a diverse U.S. AI stack.
Regulatory Outlook
While some policymakers and industry have called for aggressive restriction of open models, these efforts failed because of concerns that these measures may stifle AI innovation.
Even as policymakers remain divided on the mechanisms for domestic open-source regulation, the Trump Administration is considering a strategy to restrict U.S. companies from adopting Chinese open models. Some of the measures include adding open-source Chinese AI labs to the Department of Commerce's "Entity List" - a trade restriction list that identifies foreign companies who act against U.S. national security goals - which would effectively cut off U.S. access to Chinese technology without a license. The administration is also considering issuing government advisories on the threats of Chinese AI labs, and public pressure campaigns aimed at U.S. companies that use Chinese models. Additionally, there is growing speculation that the administration could issue an executive order on open-source AI, which could potentially limit governmental and agency use of various open-source tools.
As policymakers continue to debate how to deal with open-source AI, industry has responded in support of it. More than 20 companies signed a letter warning policymakers about the risks of "premature restrictions" on open models, arguing that those measures would reduce competition and drive innovation abroad.
While the letter does not fully address the concerns associated with open source, the debate highlights the need for a more evidence-based approach to assessing the risks associated with these models. As evidence of the incremental risks posed by open models remains limited - and in some cases speculative - policymakers should weigh the risks against the demonstrated benefits of open source for innovation, access to AI, competition, and data control.
Conclusion
As open models become more capable and widely adopted, market evolution shows that the full spectrum of AI model releases, from closed to open, brings benefits to AI development and adoption, and policymakers should weigh any potential risks against the benefits. Rather than treating open models as a problem, an evidence-based approach to regulating AI could preserve the benefits of a diverse AI stack while addressing the specific risks that come with increasingly capable models.
* * *
Original text here: https://www.americanactionforum.org/insight/rethinking-the-open-source-ai-debate/
[Category: Think Tank]
America First Policy Institute: Small Businesses Need Reliable, Affordable Energy
WASHINGTON, Sept. 3 -- The America First Policy Institute issued the following news release on Sept. 2, 2026:
* * *
Small Businesses Need Reliable, Affordable Energy
Today, the America First Policy Institute's (AFPI) Ted Ellis, deputy director for Energy and Environment Policy, testified before the House Committee on Small Business Subcommittee on Rural Development, Energy, and Supply Chains, on the need to power American businesses.
"America's small businesses depend on affordable and reliable electricity. But too many years of policies that placed reliability last, plus rapidly growing demand,
... Show Full Article
WASHINGTON, Sept. 3 -- The America First Policy Institute issued the following news release on Sept. 2, 2026:
* * *
Small Businesses Need Reliable, Affordable Energy
Today, the America First Policy Institute's (AFPI) Ted Ellis, deputy director for Energy and Environment Policy, testified before the House Committee on Small Business Subcommittee on Rural Development, Energy, and Supply Chains, on the need to power American businesses.
"America's small businesses depend on affordable and reliable electricity. But too many years of policies that placed reliability last, plus rapidly growing demand,have increased pressure on the electric grid," said Ted Ellis.
"To power America's next era of growth, we must keep reliable power plants online, build new energy resources at speed and scale, reform outdated permitting processes, and ensure small businesses and families are not left paying the costs of infrastructure built for large new electricity users."
As electricity demand continues to grow, AFPI will continue to support policies that unleash American energy dominance by strengthening the grid, expanding reliable and affordable energy, and creating new opportunities for small businesses across construction, manufacturing, and the skilled trades.
* * *
Original text here: https://www.americafirstpolicy.com/issues/small-businesses-need-reliable-affordable-energy
[Category: ThinkTank]
AFPI Releases 'Back the Blue,' a Policy Roadmap to Strengthen America's Law Enforcement Workforce
WASHINGTON, Sept. 3 -- The America First Policy Institute issued the following news release on Sept. 2, 2026:
* * *
AFPI Releases "Back the Blue," a Policy Roadmap to Strengthen America's Law Enforcement Workforce
Today, the America First Policy Institute (AFPI) released Back the Blue: A Policy Roadmap for Better-Staffed, Better-Supported Law Enforcement, a new policy paper examining America's law enforcement staffing crisis and outlining solutions for recruiting, retaining, and supporting law enforcement officers around the country.
The paper addresses the looming crisis of law enforcement
... Show Full Article
WASHINGTON, Sept. 3 -- The America First Policy Institute issued the following news release on Sept. 2, 2026:
* * *
AFPI Releases "Back the Blue," a Policy Roadmap to Strengthen America's Law Enforcement Workforce
Today, the America First Policy Institute (AFPI) released Back the Blue: A Policy Roadmap for Better-Staffed, Better-Supported Law Enforcement, a new policy paper examining America's law enforcement staffing crisis and outlining solutions for recruiting, retaining, and supporting law enforcement officers around the country.
The paper addresses the looming crisis of law enforcementofficer shortages, highlighting data showing how police departments across the country are facing major staffing shortfalls driven by retirements, resignations, recruitment challenges, burnout, compensation disparities, and burdensome hiring processes.
Despite record-low levels of crime under President Trump's leadership, officer vacancies present a major public safety concern. AFPI's 'Back the Blue' policy paper outlines suggestions for policymakers to ensure safe and protected communities.
'Back the Blue' emphasizes that supporting law enforcement is not simply about hiring more officers--it's about building stable, professional, experienced departments that are able to properly keep communities safe.
AFPI's Director of American Justice Policy Greg Glod weighs in on the recommendations included in this paper:
"America's law enforcement officers have carried a heavier load than ever over the past five years, and the recent drops in violent crime are a testament to their work. This paper gives policymakers at every level a concrete roadmap to meet that effort with the pay, staffing, and support this profession has earned. The recommendations in the paper are built on the strategies states and cities are already using to fill their ranks and drive crime down."
Read the full policy paper: Back the Blue: A Policy Roadmap for Better-Staffed, Better-Supported Law Enforcement here (https://www.americafirstpolicy.com/issues/back-the-blue-a-policy-roadmap-for-better-staffed-better-supported-law-enforcement).
* * *
Original text here: https://www.americafirstpolicy.com/issues/afpi-releases-back-the-blue-a-policy-roadmap-to-strengthen-americas-law-enforcement-workforce
[Category: ThinkTank]
