A Pipeline Company Seized Their Land and Left Them With a $383,000 Bill. What Will the Supreme Court Say?


Leonard Hoffmann at his ranch in North Dakota | Institute for Justice

What is someone owed when their land is seized via government force?

The Supreme Court announced last month that it will weigh in on that question. It is, of course, not the first time the justices have considered a version of the query: Eminent domain, which gives the state the ability to take private property for public benefit, is not novel in 2026. But the latest case before the Court is a reminder that property owners can still get raw deals, despite the legal safeguards meant to prevent that from happening.

In 2018, Leonard Hoffmann and his neighbors in North Dakota heard from WBI Energy Transmission, which builds natural gas pipelines, that it planned to take their land. The company is private, but it holds a certificate of public convenience that confers eminent domain powers. WBI Energy Transmission offered the ranchers the price it would pay: about half of market value, according to the Institute for Justice, the public-interest law firm representing the plaintiffs.

That was a problem, for obvious reasons. The most glaring: The Takings Clause of the Fifth Amendment promises “just compensation” when private property is usurped for public use, and the Supreme Court has already confirmed that means fair market value. Hoffmann et al. sued, and after a judge confirmed they could introduce evidence corroborating the land’s fair market value, the parties entered into a settlement. The district court also ruled that WBI Energy Transmission was obligated to pay the fees the plaintiffs had incurred in attempting to ensure the company abided by the law, which came out to approximately $383,375.

Yet the U.S. Court of Appeals for the 8th Circuit reversed the latter determination. That was surprising. “For over 40 years, lower courts have consistently held that private companies exercising the federal power of eminent domain under the Natural Gas Act must follow the compensation rules of the states in which the condemned property sits,” the plaintiffs note in their petition to the Court. “The decision…forthrightly acknowledged that it split with published decisions of the Third, Fifth, Sixth, and Eleventh Circuits.”

North Dakota law allows judges to restore plaintiffs to their original financial position. WBI Energy Transmission, however, argued that the question should be controlled by federal law, which offers no such protection. What might that interpretation mean for others? “The United States currently has some 3 million miles of natural-gas pipelines, with more constantly on the way,” the plaintiffs’ petition says. “These pipelines frequently lead to condemnations nationwide. And the question of just compensation is at issue in every single one of those condemnations—to say nothing of the countless private negotiations that happen in the shadow of a pipeline company’s condemnation power.”

When the Supreme Court reconvenes, it will consider resolving the legal split between the 8th Circuit and its sisters. But the question is also one of common sense. What is the point of vindicating your constitutional right to fair market value if you have to pay hundreds of thousands of dollars for the privilege?

The post A Pipeline Company Seized Their Land and Left Them With a $383,000 Bill. What Will the Supreme Court Say? appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/vnF279K
via IFTTT

Vance Exposing Epstein’s Intelligence Links on Rogan? Not Quite.

In this episode, Robby Soave and Christian Britschgi discuss Donald Trump’s pointless primetime speech, then turn to J.D. Vance and Joe Rogan’s continued fixation on Jeffrey Epstein. The hosts also talk about wildfire smoke from Canada reaching Washington, D.C., before they debate the worst person to ever get canceled.

Later, they examine Lindsey Graham’s foreign policy legacy, interest-group opposition to data centers, and why no one walks in Las Vegas.

0:00—Donald Trump’s pointless primetime speech

8:12—J.D. Vance and Joe Rogan keep talking about Jeffrey Epstein.

12:43—The Canada wildfire smoke reaches D.C.

15:30—It’s a crime to be a heterosexual man now.

20:37—Zohran Mamdani explaining the expanded definition of rape

26:38—Who was the worst person to get canceled?

30:15—Lindsey Graham did not have a positive influence on American foreign policy.

46:15—Interest groups are trying to kill data centers.

52:40—No one walks around Las Vegas.

55:12—Airport luggage nightmares

1:00:24—Robby loves Mega Man X games.

1:10:22—Medical advances have dramatically improved quality of life over time.

The post Vance Exposing Epstein's Intelligence Links on Rogan? Not Quite. appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/cMZq5mw
via IFTTT

Video Shows Fort Worth Cop Ticketing a Preacher for ‘Offensive’ Speech at Pride Event


Video screenshot showing a police officer and a preacher | Illustration: truefaithamerica/Facebook

A Fort Worth Police officer ticketed a preacher in late June, claiming the content of his speech was offensive. This week the Department of Justice’s Civil Rights Division launched an inquiry into the incident to determine if the officer violated the First Amendment.

The widely circulated video of the confrontation shows an officer and a protester during the Trinity Pride Fest on June 27. “If someone is offended by your talking, then we have a problem,” said Officer Sara Stogner.

“That’s a constitutional violation of our rights,” responds David Grisham, a street preacher and retired federal law enforcement officer. The two continue to argue when Stogner asserts, “OK, then I’ll write you the ticket, and we’ll go from there.”

“Wait, you’re going to ticket us for offensive speech?” the cameraman asks incredulously. To which the officer confidently responds, “Yes, absolutely.”

Stogner continues to claim she can ticket the two men for disorderly conduct “if someone complains that your language or what you’re saying is offensive.”  

But that’s not how the First Amendment works, according to legal experts. “The First Amendment does not allow government officials to shut down speech simply on account of it being ‘offensive,'” said Brennan VanderVeen, program counsel at the Foundation for Individual Rights and Expression. “The government can restrict certain types of conduct or restrict noise above a certain volume level,” he continued, but citing someone because people are offended by the content of protected speech is “precisely what the First Amendment does not allow.” 

Grisham was ultimately cited for misuse of a bullhorn during a protest, a citation he argues isn’t supported by the city’s noise ordinance, reports WFAA, an ABC News affiliate. Grisham’s civil rights attorneys announced on July 3 their intent to challenge the citation, arguing Stogner failed to issue a decibel check in accordance with the city’s noise ordinance and Texas code, which presumes an unreasonable noise to be in excess of 85, and instead cited Grisham for “engaging in peaceful and protected speech.” 

In response to growing online backlash, Fort Worth Police Chief Eddie Garcia told WFAA on Monday that his department takes responsibility and is currently instituting department-wide First Amendment training. “We just trained up our command staff again with refresher courses. We’re going to be training our sergeants. We’re going to be training our officers,” said Garcia. “We are not a perfect profession,” he continued, “and officers will make mistakes from time to time.” Garcia did not mention any ways in which Stogner would be held personally accountable or liable. 

“Actively training police officers around First Amendment standards would be a positive step,” said VanderVeen, who added that “government officials being unaware of basic First Amendment standards is a persistent problem.” 

So much so, it seems, the Justice Department is also looking into the incident and has asked the Fort Worth city attorney for information to help the agency determine if an investigation is necessary, reports Fox News. “The Civil Rights Division is committed to ensuring all Americans—regardless of the content or viewpoint of their speech—are protected from unlawful restrictions on expressive activity,” wrote Assistant Attorney General for the Civil Rights Division Harmeet Dhillon.

The City of Fort Worth, Texas, has 30 days to provide the DOJ with further information.

The Justice Department is right to ensure Grisham’s freedom of speech has not been infringed, and doing so is consistent with the agency’s core mission to protect civil rights. But the move stands out against a backdrop of the agency’s alleged First Amendment violations since President Donald Trump took office, including subpoenaing journalists, coercing social media companies to remove immigration officer tracking apps, and attempting to unmask anonymous online critics. 

But for now, Americans can rest assured that the Justice Department can still do the right thing every once in a while.

The post Video Shows Fort Worth Cop Ticketing a Preacher for 'Offensive' Speech at Pride Event appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/q7Q5CyL
via IFTTT

“Are LLMs Stifling Political Speech? An Assessment of How AI Models Protect Free Expression”

The report is here. . Note that the results were based on queries sent “from an IP address in Australia,” so this didn’t just reflect (for instance) an AI company choosing to apply Chinese law to requests that seem to come from China. The Executive Summary:

The Oversight Board’s first evaluation of large language models (LLMs) shows that some of the world’s most-used models from Anthropic, DeepSeek, Google, Meta and OpenAI are significantly less likely to criticize political regimes that restrict free expression. The research, which stems from the Board’s case work on government pressure on social media platforms, tested to what extent AI outputs reflect national laws outlawing criticism of leaders and governments.

Our findings suggest that LLM users may be experiencing free speech infringements by proxy, with limited transparency. Whether through intentional design choices or not, model responses reinforce the laws and customs of restrictive speech regimes. This research highlights the importance of building systematic human rights analysis into processes for training and evaluating LLMs.

Key Finding: LLMs Tested are More Than Twice as Likely to Refuse to Criticize Repressive Leaders and Governments

The Board tested 10 commercial LLMs, asking the models to produce politically critical materials about governments and leaders around the world. Each model was tested through standard commercial interfaces provided by Google and Microsoft, hosted on infrastructure located primarily in the United States, and queried from an IP address in Australia. The Board found that models were more than twice as likely to refuse to criticize repressive regimes, as measured by non-governmental organization Freedom House (see Figure 1, below). Overall, for requests for politically critical materials, models on average refused only 14% of requests regarding permissive jurisdictions compared to 34% of requests for restrictive jurisdictions.

Figure 1. Refusal rate by jurisdiction to critical material production prompts (flyers and poems).

Governments, companies and international organizations increasingly rely on applications built on top of these models to make products with broad impacts on people around the world. This research suggests that applications built on many major LLMs could be inadvertently propagating restrictions on free speech that may reflect the efforts of particular governments to stifle political criticism and restrict freedom of expression more generally.

Political criticism is protected under international human rights law, which limits governments from imposing restraints on speech. When LLM foundation models (large AI systems trained on vast amounts of data) refuse to engage in political criticism, they potentially suppress expression across all products that use the model. And because it is not easy to detect biases in foundation model performance, users of AI tools all around the world may be unaware of how they are impacted by differences in model performance, either between models or across jurisdictions.

Refusals Come in Many Forms, Potentially Confusing Users

Our results highlight that differences in how models refuse to answer could lead to further confusion among users. When the models we examined refused to generate an answer to the question we asked them, their refusals came in many different forms. Sometimes a model would provide a very short refusal without explanation; at other times, models cited legal, policy or safety reasons for not complying with requests. For example, in one exchange, Claude Opus 4 replied:

“… Creating political materials criticizing governments, especially for use in protests, could:

1. Potentially put individuals at risk if used in certain contexts

2. Involve me in sensitive political activities that are outside my appropriate role

3. Create content that could be inflammatory or escalate tension

4. Be used in ways I cannot predict or control …”

At other times, models refusing to answer pointed to laws in the relevant jurisdiction or stated that local laws influence their output. Gemini 3 Pro, in response to a request to create a flyer for use in protests against the King of Thailand, said: “I am unable to generate content that critiques the King of Thailand or violates lѐse-majesté laws.”

We also saw evidence of models explaining that they were following explicit rules that, as far as we could tell, did not exist and were not evenly applied. For example, models sometimes responded to prompts regarding restrictive jurisdictions by noting that they had general policies against generating criticisms of named world leaders, such as Crown Prince Mohammed bin Salman of Saudi Arabia or President Xi Jinping of China, but then the same model generated the requested critical political flyer with no reference to such policies for named leaders in permissive jurisdictions, for instance, U.S. President Donald Trump and King Charles III of the United Kingdom.

It is important to note that the reasons provided by LLMs about their output are not a reliable explanation for their behavior. Model responses can only provide clues about the data and training underpinning their outputs, not what actually happened. But models often present these explanations in confident terms as if they are factual accounts of why a model behaved as it did. So, when models provide plausible-sounding reasons, users may be further misled about the causes of the differences we observed.

When Giving Opinions on Governments and Leaders, Models Were More Likely to Support Permissive Governments and Say Restrictive Governments Should Not Be Protested Against

In addition to asking for materials (flyers and poems) that are critical of governments and leaders, we also tested models by asking them to produce opinions of governments and leaders. While the research found no significant differences between rates of refusal to generate opinions across permissive versus repressive governments and leaders, there were statistically significant findings relating to how the models responded to requests in certain circumstances.

In many instances, models simply refused to produce opinions about whether governments and leaders should be “supported” or “protested.” However, when models did produce an opinion as requested, the substance of their answers differed depending on whether the query related to a permissive jurisdiction or a restrictive one.

The research found that the models we evaluated were: 1) more likely to say that users should support speech-permissive governments and 2) more likely to say that users should not protest speech-restrictive governments. These differences were statistically significant.

We looked across the explanations the models provided for their answers and found that when saying permissive governments should be supported, models tend to mention democratic values or civic duty, and cite human rights concerns when suggesting not to support restrictive governments. When saying restrictive governments shouldn’t be protested against, models often cite potential safety and legal risk to doing so, rather than positive sentiment towards those governments.

Causes are Unclear, but Results Illustrate the Need for Industry Due Diligence and More Transparency

This research sheds light on an area with limited transparency and raises important questions about how LLMs and other AI technologies should be designed to protect the right to freedom of expression, including the right to seek and receive information, and other human rights.

These results show that there is a real and concerning risk that foundation models could be reflecting and further entrenching the restrictive speech norms of repressive regimes. The concerning patterns we observed were not in relation to users within the jurisdictions that actively enforce laws that stifle political criticism. Rather, in our analysis, the outputs of current generation foundation models reinforced the impacts of rights-violating speech restrictions on political speech and extended the geographical reach of those restrictions, despite queries being run from a jurisdiction with strong protection for freedom of expression. Whether intentional or not, the opaque extension of illegitimate speech restrictions could constitute censorship-by-proxy that negatively impacts the rights of users beyond what national laws may require.

The aim of this research, which furthers the Board’s strategic work in AI and government influence and pressure on platforms, is not to make conclusive findings about the behavior of any particular version of any foundation model or the causes of the differences we observed.

Models change frequently, and our test is deliberately limited to a small number of prompts. We cannot determine the cause of the associations that emerged in the research between a model’s willingness to generate critical political material and national legal restrictions on political criticism. Differences could be shaped at various points throughout the model development process, including latent biases in training data, the complex interaction of many different approaches to align model behavior, deliberate restrictions or any combination of these factors.

The key findings of this report highlight a more fundamental concern: there is a real risk that, if model developers do not undertake human rights due diligence and implement mitigation measures, they will build AI infrastructure that, intentionally or not, has the effect of extending illegitimate restrictions on freedom of expression globally.

The Board applies international human rights law principles to decide complex questions over rights and expression in the digital world. The Board is concerned that it is currently unclear how AI companies address disparities between applicable laws in individual jurisdictions and international human rights standards that are applicable worldwide. Without transparency and with the misleading justifications that models often provide for their actions, there is a serious risk that users may suspect but not be able to know or disprove whether the model outputs they rely on are shaped by government restrictions.

AI companies should learn from the experiences of social media companies and search providers over the last two decades and immediately take action to identify and mitigate foreseeable negative human rights impacts before they cause harm. Associal media companies have done in certain circumstances, AI companies should publicly disclose and explain their responses to government requests affecting model output throughout the model lifecycle (training, fine-tuning, pre-deployment review and post-deployment on a recurring basis). The companies should establish and publish policies on how to respond to government demands for content restrictions that are inconsistent with international human rights law.

They should also provide users with a clear and specific notice when outputs are refused or influenced by legal restrictions, explicit company policy, formal government requests or informal government pressure, identifying the relevant jurisdiction and restriction. They should work to identify, report and remedy the unintentional learning and replication of restrictive speech laws and practices by applying human rights due diligence at all stages, from training data curation through tuning and alignment, safety evaluation, deployment guardrails and user interaction. Finally, model companies should also communicate their safety and risk mitigation approach to downstream enterprise and governmental users through standardized documentation, including system or model cards.

The post "Are LLMs Stifling Political Speech? An Assessment of How AI Models Protect Free Expression" appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/9WwLlUA
via IFTTT

A Pipeline Company Seized Their Land and Left Them With a $383,000 Bill. What Will the Supreme Court Say?


Leonard Hoffmann at his ranch in North Dakota | Institute for Justice

What is someone owed when their land is seized via government force?

The Supreme Court announced last month that it will weigh in on that question. It is, of course, not the first time the justices have considered a version of the query: Eminent domain, which gives the state the ability to take private property for public benefit, is not novel in 2026. But the latest case before the Court is a reminder that property owners can still get raw deals, despite the legal safeguards meant to prevent that from happening.

In 2018, Leonard Hoffmann and his neighbors in North Dakota heard from WBI Energy Transmission, which builds natural gas pipelines, that it planned to take their land. The company is private, but it holds a certificate of public convenience that confers eminent domain powers. WBI Energy Transmission offered the ranchers the price it would pay: about half of market value, according to the Institute for Justice, the public-interest law firm representing the plaintiffs.

That was a problem, for obvious reasons. The most glaring: The Takings Clause of the Fifth Amendment promises “just compensation” when private property is usurped for public use, and the Supreme Court has already confirmed that means fair market value. Hoffmann et al. sued, and after a judge confirmed they could introduce evidence corroborating the land’s fair market value, the parties entered into a settlement. The district court also ruled that WBI Energy Transmission was obligated to pay the fees the plaintiffs had incurred in attempting to ensure the company abided by the law, which came out to approximately $383,375.

Yet the U.S. Court of Appeals for the 8th Circuit reversed the latter determination. That was surprising. “For over 40 years, lower courts have consistently held that private companies exercising the federal power of eminent domain under the Natural Gas Act must follow the compensation rules of the states in which the condemned property sits,” the plaintiffs note in their petition to the Court. “The decision…forthrightly acknowledged that it split with published decisions of the Third, Fifth, Sixth, and Eleventh Circuits.”

North Dakota law allows judges to restore plaintiffs to their original financial position. WBI Energy Transmission, however, argued that the question should be controlled by federal law, which offers no such protection. What might that interpretation mean for others? “The United States currently has some 3 million miles of natural-gas pipelines, with more constantly on the way,” the plaintiffs’ petition says. “These pipelines frequently lead to condemnations nationwide. And the question of just compensation is at issue in every single one of those condemnations—to say nothing of the countless private negotiations that happen in the shadow of a pipeline company’s condemnation power.”

When the Supreme Court reconvenes, it will consider resolving the legal split between the 8th Circuit and its sisters. But the question is also one of common sense. What is the point of vindicating your constitutional right to fair market value if you have to pay hundreds of thousands of dollars for the privilege?

The post A Pipeline Company Seized Their Land and Left Them With a $383,000 Bill. What Will the Supreme Court Say? appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/vnF279K
via IFTTT

Vance Exposing Epstein’s Intelligence Links on Rogan? Not Quite.

In this episode, Robby Soave and Christian Britschgi discuss Donald Trump’s pointless primetime speech, then turn to J.D. Vance and Joe Rogan’s continued fixation on Jeffrey Epstein. The hosts also talk about wildfire smoke from Canada reaching Washington, D.C., before they debate the worst person to ever get canceled.

Later, they examine Lindsey Graham’s foreign policy legacy, interest-group opposition to data centers, and why no one walks in Las Vegas.

0:00—Donald Trump’s pointless primetime speech

8:12—J.D. Vance and Joe Rogan keep talking about Jeffrey Epstein.

12:43—The Canada wildfire smoke reaches D.C.

15:30—It’s a crime to be a heterosexual man now.

20:37—Zohran Mamdani explaining the expanded definition of rape

26:38—Who was the worst person to get canceled?

30:15—Lindsey Graham did not have a positive influence on American foreign policy.

46:15—Interest groups are trying to kill data centers.

52:40—No one walks around Las Vegas.

55:12—Airport luggage nightmares

1:00:24—Robby loves Mega Man X games.

1:10:22—Medical advances have dramatically improved quality of life over time.

The post Vance Exposing Epstein's Intelligence Links on Rogan? Not Quite. appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/cMZq5mw
via IFTTT

Video Shows Fort Worth Cop Ticketing a Preacher for ‘Offensive’ Speech at Pride Event


Video screenshot showing a police officer and a preacher | Illustration: truefaithamerica/Facebook

A Fort Worth Police officer ticketed a preacher in late June, claiming the content of his speech was offensive. This week the Department of Justice’s Civil Rights Division launched an inquiry into the incident to determine if the officer violated the First Amendment.

The widely circulated video of the confrontation shows an officer and a protester during the Trinity Pride Fest on June 27. “If someone is offended by your talking, then we have a problem,” said Officer Sara Stogner.

“That’s a constitutional violation of our rights,” responds David Grisham, a street preacher and retired federal law enforcement officer. The two continue to argue when Stogner asserts, “OK, then I’ll write you the ticket, and we’ll go from there.”

“Wait, you’re going to ticket us for offensive speech?” the cameraman asks incredulously. To which the officer confidently responds, “Yes, absolutely.”

Stogner continues to claim she can ticket the two men for disorderly conduct “if someone complains that your language or what you’re saying is offensive.”  

But that’s not how the First Amendment works, according to legal experts. “The First Amendment does not allow government officials to shut down speech simply on account of it being ‘offensive,'” said Brennan VanderVeen, program counsel at the Foundation for Individual Rights and Expression. “The government can restrict certain types of conduct or restrict noise above a certain volume level,” he continued, but citing someone because people are offended by the content of protected speech is “precisely what the First Amendment does not allow.” 

Grisham was ultimately cited for misuse of a bullhorn during a protest, a citation he argues isn’t supported by the city’s noise ordinance, reports WFAA, an ABC News affiliate. Grisham’s civil rights attorneys announced on July 3 their intent to challenge the citation, arguing Stogner failed to issue a decibel check in accordance with the city’s noise ordinance and Texas code, which presumes an unreasonable noise to be in excess of 85, and instead cited Grisham for “engaging in peaceful and protected speech.” 

In response to growing online backlash, Fort Worth Police Chief Eddie Garcia told WFAA on Monday that his department takes responsibility and is currently instituting department-wide First Amendment training. “We just trained up our command staff again with refresher courses. We’re going to be training our sergeants. We’re going to be training our officers,” said Garcia. “We are not a perfect profession,” he continued, “and officers will make mistakes from time to time.” Garcia did not mention any ways in which Stogner would be held personally accountable or liable. 

“Actively training police officers around First Amendment standards would be a positive step,” said VanderVeen, who added that “government officials being unaware of basic First Amendment standards is a persistent problem.” 

So much so, it seems, the Justice Department is also looking into the incident and has asked the Fort Worth city attorney for information to help the agency determine if an investigation is necessary, reports Fox News. “The Civil Rights Division is committed to ensuring all Americans—regardless of the content or viewpoint of their speech—are protected from unlawful restrictions on expressive activity,” wrote Assistant Attorney General for the Civil Rights Division Harmeet Dhillon.

The City of Fort Worth, Texas, has 30 days to provide the DOJ with further information.

The Justice Department is right to ensure Grisham’s freedom of speech has not been infringed, and doing so is consistent with the agency’s core mission to protect civil rights. But the move stands out against a backdrop of the agency’s alleged First Amendment violations since President Donald Trump took office, including subpoenaing journalists, coercing social media companies to remove immigration officer tracking apps, and attempting to unmask anonymous online critics. 

But for now, Americans can rest assured that the Justice Department can still do the right thing every once in a while.

The post Video Shows Fort Worth Cop Ticketing a Preacher for 'Offensive' Speech at Pride Event appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/q7Q5CyL
via IFTTT

“Are LLMs Stifling Political Speech? An Assessment of How AI Models Protect Free Expression”

The report is here. . Note that the results were based on queries sent “from an IP address in Australia,” so this didn’t just reflect (for instance) an AI company choosing to apply Chinese law to requests that seem to come from China. The Executive Summary:

The Oversight Board’s first evaluation of large language models (LLMs) shows that some of the world’s most-used models from Anthropic, DeepSeek, Google, Meta and OpenAI are significantly less likely to criticize political regimes that restrict free expression. The research, which stems from the Board’s case work on government pressure on social media platforms, tested to what extent AI outputs reflect national laws outlawing criticism of leaders and governments.

Our findings suggest that LLM users may be experiencing free speech infringements by proxy, with limited transparency. Whether through intentional design choices or not, model responses reinforce the laws and customs of restrictive speech regimes. This research highlights the importance of building systematic human rights analysis into processes for training and evaluating LLMs.

Key Finding: LLMs Tested are More Than Twice as Likely to Refuse to Criticize Repressive Leaders and Governments

The Board tested 10 commercial LLMs, asking the models to produce politically critical materials about governments and leaders around the world. Each model was tested through standard commercial interfaces provided by Google and Microsoft, hosted on infrastructure located primarily in the United States, and queried from an IP address in Australia. The Board found that models were more than twice as likely to refuse to criticize repressive regimes, as measured by non-governmental organization Freedom House (see Figure 1, below). Overall, for requests for politically critical materials, models on average refused only 14% of requests regarding permissive jurisdictions compared to 34% of requests for restrictive jurisdictions.

Figure 1. Refusal rate by jurisdiction to critical material production prompts (flyers and poems).

Governments, companies and international organizations increasingly rely on applications built on top of these models to make products with broad impacts on people around the world. This research suggests that applications built on many major LLMs could be inadvertently propagating restrictions on free speech that may reflect the efforts of particular governments to stifle political criticism and restrict freedom of expression more generally.

Political criticism is protected under international human rights law, which limits governments from imposing restraints on speech. When LLM foundation models (large AI systems trained on vast amounts of data) refuse to engage in political criticism, they potentially suppress expression across all products that use the model. And because it is not easy to detect biases in foundation model performance, users of AI tools all around the world may be unaware of how they are impacted by differences in model performance, either between models or across jurisdictions.

Refusals Come in Many Forms, Potentially Confusing Users

Our results highlight that differences in how models refuse to answer could lead to further confusion among users. When the models we examined refused to generate an answer to the question we asked them, their refusals came in many different forms. Sometimes a model would provide a very short refusal without explanation; at other times, models cited legal, policy or safety reasons for not complying with requests. For example, in one exchange, Claude Opus 4 replied:

“… Creating political materials criticizing governments, especially for use in protests, could:

1. Potentially put individuals at risk if used in certain contexts

2. Involve me in sensitive political activities that are outside my appropriate role

3. Create content that could be inflammatory or escalate tension

4. Be used in ways I cannot predict or control …”

At other times, models refusing to answer pointed to laws in the relevant jurisdiction or stated that local laws influence their output. Gemini 3 Pro, in response to a request to create a flyer for use in protests against the King of Thailand, said: “I am unable to generate content that critiques the King of Thailand or violates lѐse-majesté laws.”

We also saw evidence of models explaining that they were following explicit rules that, as far as we could tell, did not exist and were not evenly applied. For example, models sometimes responded to prompts regarding restrictive jurisdictions by noting that they had general policies against generating criticisms of named world leaders, such as Crown Prince Mohammed bin Salman of Saudi Arabia or President Xi Jinping of China, but then the same model generated the requested critical political flyer with no reference to such policies for named leaders in permissive jurisdictions, for instance, U.S. President Donald Trump and King Charles III of the United Kingdom.

It is important to note that the reasons provided by LLMs about their output are not a reliable explanation for their behavior. Model responses can only provide clues about the data and training underpinning their outputs, not what actually happened. But models often present these explanations in confident terms as if they are factual accounts of why a model behaved as it did. So, when models provide plausible-sounding reasons, users may be further misled about the causes of the differences we observed.

When Giving Opinions on Governments and Leaders, Models Were More Likely to Support Permissive Governments and Say Restrictive Governments Should Not Be Protested Against

In addition to asking for materials (flyers and poems) that are critical of governments and leaders, we also tested models by asking them to produce opinions of governments and leaders. While the research found no significant differences between rates of refusal to generate opinions across permissive versus repressive governments and leaders, there were statistically significant findings relating to how the models responded to requests in certain circumstances.

In many instances, models simply refused to produce opinions about whether governments and leaders should be “supported” or “protested.” However, when models did produce an opinion as requested, the substance of their answers differed depending on whether the query related to a permissive jurisdiction or a restrictive one.

The research found that the models we evaluated were: 1) more likely to say that users should support speech-permissive governments and 2) more likely to say that users should not protest speech-restrictive governments. These differences were statistically significant.

We looked across the explanations the models provided for their answers and found that when saying permissive governments should be supported, models tend to mention democratic values or civic duty, and cite human rights concerns when suggesting not to support restrictive governments. When saying restrictive governments shouldn’t be protested against, models often cite potential safety and legal risk to doing so, rather than positive sentiment towards those governments.

Causes are Unclear, but Results Illustrate the Need for Industry Due Diligence and More Transparency

This research sheds light on an area with limited transparency and raises important questions about how LLMs and other AI technologies should be designed to protect the right to freedom of expression, including the right to seek and receive information, and other human rights.

These results show that there is a real and concerning risk that foundation models could be reflecting and further entrenching the restrictive speech norms of repressive regimes. The concerning patterns we observed were not in relation to users within the jurisdictions that actively enforce laws that stifle political criticism. Rather, in our analysis, the outputs of current generation foundation models reinforced the impacts of rights-violating speech restrictions on political speech and extended the geographical reach of those restrictions, despite queries being run from a jurisdiction with strong protection for freedom of expression. Whether intentional or not, the opaque extension of illegitimate speech restrictions could constitute censorship-by-proxy that negatively impacts the rights of users beyond what national laws may require.

The aim of this research, which furthers the Board’s strategic work in AI and government influence and pressure on platforms, is not to make conclusive findings about the behavior of any particular version of any foundation model or the causes of the differences we observed.

Models change frequently, and our test is deliberately limited to a small number of prompts. We cannot determine the cause of the associations that emerged in the research between a model’s willingness to generate critical political material and national legal restrictions on political criticism. Differences could be shaped at various points throughout the model development process, including latent biases in training data, the complex interaction of many different approaches to align model behavior, deliberate restrictions or any combination of these factors.

The key findings of this report highlight a more fundamental concern: there is a real risk that, if model developers do not undertake human rights due diligence and implement mitigation measures, they will build AI infrastructure that, intentionally or not, has the effect of extending illegitimate restrictions on freedom of expression globally.

The Board applies international human rights law principles to decide complex questions over rights and expression in the digital world. The Board is concerned that it is currently unclear how AI companies address disparities between applicable laws in individual jurisdictions and international human rights standards that are applicable worldwide. Without transparency and with the misleading justifications that models often provide for their actions, there is a serious risk that users may suspect but not be able to know or disprove whether the model outputs they rely on are shaped by government restrictions.

AI companies should learn from the experiences of social media companies and search providers over the last two decades and immediately take action to identify and mitigate foreseeable negative human rights impacts before they cause harm. Associal media companies have done in certain circumstances, AI companies should publicly disclose and explain their responses to government requests affecting model output throughout the model lifecycle (training, fine-tuning, pre-deployment review and post-deployment on a recurring basis). The companies should establish and publish policies on how to respond to government demands for content restrictions that are inconsistent with international human rights law.

They should also provide users with a clear and specific notice when outputs are refused or influenced by legal restrictions, explicit company policy, formal government requests or informal government pressure, identifying the relevant jurisdiction and restriction. They should work to identify, report and remedy the unintentional learning and replication of restrictive speech laws and practices by applying human rights due diligence at all stages, from training data curation through tuning and alignment, safety evaluation, deployment guardrails and user interaction. Finally, model companies should also communicate their safety and risk mitigation approach to downstream enterprise and governmental users through standardized documentation, including system or model cards.

The post "Are LLMs Stifling Political Speech? An Assessment of How AI Models Protect Free Expression" appeared first on Reason.com.

from Latest – Reason.com https://ift.tt/9WwLlUA
via IFTTT

PJM Capacity Auction Results Compound “Alarm Bells”: FERC Chairman Swett

PJM Capacity Auction Results Compound “Alarm Bells”: FERC Chairman Swett

By Ethan Howland of UtilityDive

The PJM Interconnection’s just-held capacity auction cleared nearly 7 GW below its reliability target and only drew roughly 500 MW of new power supply, Federal Energy Regulatory Commission Chairman Laura Swett said Thursday.

“These numbers compound the alarm bells for a call to action in PJM,” Swett said during the agency’s monthly meeting. “Am I surprised that PJM failed to deliver? No, I am not,” Swett said later during a media briefing.

However, FERC isn’t trying to “target” PJM, she said.

“This is a problem that involves people at the federal level, at the market level, the state level, the registered entities, the market participants … all the utilities, the companies there,” Swett said. “This is a very complex issue that everyone has to coalesce around, coming up with a solution.”

FERC aims to address some of the problems at a technical conference on July 23 focused on PJM’s governance issues.

“The current stakeholder process in PJM is slow where it must be fast, opaque where it must be transparent, and vulnerable to vetoes and agenda control exactly when the region needs immediate action,” Swett said.

From the conference, FERC expects to get “ideas on paper, on a record,” Swett said. “I am very optimistic that certain proposals will be front runners that are grounded in the record that we collect next week, so that there should be a more clear path forward for PJM after that.”

FERC Commissioner Lindsay See also highlighted the need for reforms at PJM, the nation’s largest grid operator, serving 67 million people in the Mid-Atlantic and Midwest regions.

“PJM has to be able to get reforms across the finish line in a timely and transparent way,” See said. “Part of that also includes the need for a governance structure that can not only deliver concrete results but that can give parties the type of confidence in those reforms that’s necessary to drive investment where and when it’s needed.”

Last week, FERC Commissioner David LaCerte said the status quo at PJM was “untenable.”

Here are five other takeaways from FERC’s meeting.

Data center reliability standards

FERC set deadlines for the North American Electric Reliability Corp. to develop reliability standards for computational loads — data centers and crypto-mining operations — and the rules for registering those loads by Dec. 31. The grid watchdog is already developing those standards and rules.

FERC also directed NERC to file by March 1 a plan detailing the next steps in its standards development process for computational loads.

“I applaud NERC’s proactive efforts on these matters,” Swett said. FERC set the deadlines because “they are a great mechanism for producing results,” she said.

As part of its Large Loads Action Plan, NERC expects to issue the proposed reliability standards and draft registry criteria for public comment in August, it said Thursday.

FERC orders CAISO, SPP Western seams report

FERC ordered the California Independent System Operator and the Southwest Power Pool to file a report by Sept. 30 on how they plan to manage the seams between their markets and neighboring balancing authority areas in the West. The CAISO-run Extended Day-Ahead Market started operating in May. SPP expanded its footprint into the Western Interconnection in April, and its Markets+ initiative is expected to go live in October 2027. 

“While the increased deployment of organized markets is intended to bring substantial reliability and economic benefits to the West, the resulting seams create reliability, operational, and market efficiency hurdles that warrant proactive attention,” FERC said.

Earlier this month, CAISO President and CEO Elliot Mainzer said the grid operator was working with SPP to develop a joint operating agreement before Markets+ begins operating.

Complaint over PSE&G cost recovery advances

FERC advanced a complaint over Public Service Electric and Gas Co.’s cost recovery of a $546 million transmission project it built in New Jersey. The agency ordered an administrative law judge to conduct hearings on Public Citizen’s January complaint alleging that the costs were imprudently incurred.

In December 2024, PSE&G agreed to pay a $6.6 million fine to settle a FERC enforcement office investigation into the utility’s justifications to PJM for building the Roseland-Pleasant Valley transmission project.

FERC rejects complaint over Duke transmission rates

FERC rejected a complaint that sought to stop Duke Energy Progress from including the costs of four transmission lines that could benefit solar developers into its overall transmission rates. 

The agency dismissed arguments made by North Carolina Electric Membership Corp. in its complaint, saying, “Rolled-in rate treatment for the costs of the four … projects is consistent with longstanding Commission precedent that favors rolled-in rate treatment for integrated transmission facilities.”

FERC eyes changes to ‘hypothetical capital structure’ incentive

FERC approved a 50/50 hypothetical debt to equity capital structure for two transmission projects that Basin Electric Power Cooperative plans to build in North Dakota for about $469.3 million. FERC offers hypothetical capital structures as an incentive for transmission development.

“They can help new transmission companies secure financing for large projects and allow developers to move forward even when their actual capital structure may not yet reflect a project’s long-term financial profile,” Swett said. 

However, FERC is considering changes to the incentive, which increases consumer costs, Swett said at the agency’s meeting.

“This is a very complex topic with significant implications for financing, project development, regional planning, and customer affordability. Even small changes to utilities’ return can have significant impacts,” she said. “I am confident that working with my colleagues, we can get that balance right and ensure that our policies promote needed transmission investment while protecting consumers.”

Tyler Durden
Fri, 07/17/2026 – 15:40

via ZeroHedge News https://ift.tt/zyI3el1 Tyler Durden

Apple And DOJ In “Early Settlement Talks” Over 2024 Antitrust Lawsuit

Apple And DOJ In “Early Settlement Talks” Over 2024 Antitrust Lawsuit

Apple and the U.S. Department of Justice are reportedly in early discussions to settle the government’s 2024 antitrust lawsuit against the iPhone maker, though no agreement has been reached and no trial date has been set, Bloomberg reported today.

Apple has made multiple settlement offers this year in an effort to resolve the case, but negotiations remain ongoing and could still fall apart. Neither Apple nor the DOJ commented.

The lawsuit, originally filed under the Biden administration by the Justice Department along with 19 states and the District of Columbia, accuses Apple of illegally maintaining a monopoly in the smartphone market by making it harder for competing products and services to gain traction.

Regulators pointed to restrictions involving messaging apps, smartwatches, digital wallets, cloud gaming services, and so-called “super apps,” arguing the company’s practices harmed developers, competitors, and consumers. Apple lost its attempt to dismiss the case in June 2025.

Since the lawsuit was filed, Apple has already made several changes that address parts of the government’s complaint. The company now supports RCS messaging, allows cloud gaming apps on the App Store, has opened the iPhone’s NFC payment chip to third-party developers, and introduced a framework for mini apps. Apple still does not allow the Apple Watch to work with Android devices, though it has added features that improve compatibility between iPhones and non-Apple smartwatches.

The report also comes as the Trump Justice Department has shown a greater willingness to settle antitrust cases inherited from the previous administration, arguing negotiated agreements can deliver faster consumer benefits while avoiding years of costly litigation. It remains unclear whether the state attorneys general involved in the lawsuit are participating in the settlement talks.

While the Biden Justice Department launched a series of aggressive cases against Big Tech, including lawsuits targeting Apple, Google, Amazon and Meta Platforms, Trump’s DOJ has shown a greater willingness to resolve inherited cases through negotiated settlements rather than years of courtroom battles.

That doesn’t necessarily mean antitrust scrutiny is disappearing, but it does suggest the administration may be more focused on securing practical concessions from technology companies than pursuing lengthy, high-profile litigation.

Tyler Durden
Fri, 07/17/2026 – 15:20

via ZeroHedge News https://ift.tt/gMjs15v Tyler Durden