Donald Trump’s net approval has slumped to its lowest this term as the United States government shutdown breaks the record for the longest shutdown. Democrats had big wins in state elections on Tuesday.

I previously covered the ongoing US government shutdown on October 9, eight days into a shutdown that began on October 1. This shutdown has now lasted 38 days, beating the previous record 35-day shutdown that was set during Trump’s first term.

Although Republicans hold the presidency and majorities in both chambers of Congress, they cannot pass a budget without Democratic support in the Senate owing to the Senate’s requirement for 60 votes out of 100 senators to invoke “cloture” and end a “filibuster”.

Republicans hold a 53–47 majority over Democrats in the Senate, so they need seven Democrats to vote with them to obtain cloture. Democrats are refusing to help to pass a budget unless health insurance subsidies are extended.

For the first three weeks of the shutdown, Trump’s ratings were resilient, with his net approval in analyst Nate Silver’s aggregate of national polls rising slightly to -7.5 on October 19.

But since then, Trump’s net approval has slumped 5.5 points to -13.0, a low for him this term. Currently, 55.1% disapprove of Trump’s performance while 42.1% approve.

Trump’s net approval on the four issues tracked by Silver have all fallen recently. He now has a net approval of -4.9 on immigration, -17.6 on the economy, -17.8 on trade and -28.9 on inflation.

In Silver’s historical comparison on how Trump’s ratings compare with previous presidents since Harry Truman at this point in their presidencies, Trump’s net approval is only better than during his own first term. Joe Biden’s net approval was -8.3 at this point, making him the next worst on net approval.

Since a peak for the US benchmark S&P 500 stock market index on October 29, it has lost 2.5%. But in the last six months, it has gained nearly 20%.

Trump’s ratings will probably rebound if the shutdown ends soon. Unless something goes badly wrong with the US economy or the stock market, his ratings will probably return to net high single-digit negative, not net double-digit negative.

Democrats had big wins at state elections

US state elections occurred on Tuesday in New Jersey and Virginia. Democrats won the Virginia governorship by 57.2–42.6 over Republicans, a gain for Democrats. They also won the other two statewide races for lieutenant-governor and attorney-general.

Democrats won the lower house of the Virginia legislature by 64–36, a 13-seat gain for Democrats. The upper house was not up for election, but Democrats hold a 21–19 majority there. At the 2024 presidential election, Kamala Harris defeated Trump in Virginia by 5.8 points, though Trump won the overall popular vote by 1.5 points.

Democrats held the New Jersey governorship, winning by 56.4–43.0, far exceeding polls that gave Democrats a low single-digit lead. They lead in the lower house by 53–19, with eight seats uncalled.

If the uncalled seats go to current leaders, Democrats will win by 57–23, a five-seat gain. Democrats hold the upper house by 25–15, which was not up for election. Harris beat Trump in New Jersey in 2024 by 5.9 points.

In June, democratic socialist Zohran Mamdani had won the New York City Democratic mayoral nomination, defeating former New York governor Andrew Cuomo by 56.4–43.6 after preferences to win the Democratic primary. On Tuesday, Mamdani defeated Cuomo, who ran as an independent, in the general election
by 50.4–41.6, with 7.1% for a Republican.

Unlike the primary, the general election used first past the post. But preferences would not have changed the outcome as Mamdani exceeded 50%.

In response to Texas Republicans gerrymandering Texas to create five additional federal Republican seats, California Democrats proposed retaliatory gerrymandering of California’s federal seats. A referendum was needed to approve this gerrymander. With 79% reporting, “yes” to gerrymandering had won by 63.9–36.1. Harris won California in 2024 by 20.1 points.

See also my coverage of these elections for The Poll Bludger. In this piece, I wrote about past and upcoming elections in the Netherlands, Bolivia and Chile.

Implications for the 2026 midterm elections

At November 2026 midterm elections, all of the House of Representatives and one-third of the Senate will be up for election. In Virginia and New Jersey’s gubernatorial elections, there were respectively 8.8 and 7.5 point swings to Democrats from the 2024 presidential margin in those states.

If these swings are applied to Trump’s national margin of 1.5 points in 2024, Democrats would win nationally by 6.0 points (New Jersey swing) or 7.3 (Virginia swing). So if the swing in either state occurs nationally in 2026, Democrats are very likely to gain control of the House.

There will be 35 seats up for election in the Senate next November (33 regular and two special elections). Republicans hold 22 and Democrats 13, but only two Republican seats are thought vulnerable: Maine and North Carolina.

In 2024, Harris won Maine by 6.9 points and Trump only won North Carolina by 2.2 points. Trump won all other states Republicans are defending by at least a double-digit margin. Even if the swing in Virginia happened nationally, Democrats would gain only two seats and Republicans would hold the Senate by 51–49.

It’s become increasingly difficult for Democrats to win the Senate, as the two senators per state rule skews Senate elections towards low-population, rural states.

In the Fiftyplusone generic ballot average, Democrats lead Republicans by 45.0–41.9. The low single-digit lead for Democrats hasn’t changed since April. The current 3.1-point Democrat lead is below what happens from applying the swing in New Jersey and Virginia nationally.

While Trump’s ratings have dropped, there hasn’t been a Democratic surge on the generic ballot. That suggests voters are blaming both parties for the shutdown.

Adrian Beaumont, Election Analyst (Psephologist) at The Conversation; and Honorary Associate, School of Mathematics and Statistics, The University of Melbourne

This article is republished from The Conversation under a Creative Commons license. Read the original article.

The US Supreme Court has heard arguments overnight on the legality of President Donald Trump’s “liberation day” tariffs on most countries around the world.

The number of sceptical questions posed by the justices in the hearings was striking for a court that is dominated by conservative appointees by six to three.

At stake is not only whether the sweeping tariffs will be upheld, but the extent to which the Supreme Court is willing to extend the limits of presidential power.

So, what will the the court have to consider?

Where’s the emergency?

Trump issued these tariffs in April claiming an economic emergency, using the International Emergency Economic Powers Act (IEEPA) of 1977. So, the two primary legal questions for the court to consider are:

1. whether the IEEPA authorises Trump to issue widespread tariffs; and
2. if the IEEPA does authorise tariffs, whether it delegates authority to the president in an unconstitutional manner.

These questions have already been considered by three lower US courts, including the United States Court of International Trade. All three courts found that Trump’s tariffs were illegal.

Trump claims his power to impose tariffs is derived from the words “regulate … importation” in the IEEPA. However, justices from both sides of politics expressed scepticism about how much authority that implied. The majority in one of the lower courts described the phrase as “a wafer-thin reed”.

Supreme Court Justice Brett Kavanaugh, a Trump appointee, said:

Figuring out what ‘regulate importation’ means is – is obviously central here […] One problem you have is that presidents since IEEPA have not done this.

Chief Justice John Roberts and Justice Amy Coney Barrett, both conservatives, expressed doubt about that phrase authorising tariffs of the scale of the “liberation day” tariffs. Justice Roberts said:

The justification is being used for a power to impose tariffs on any product from any country for – in any amount for any length of time. […] that’s major authority, and the basis for the claim seems to be a misfit.

Justice Elena Kagan, a Democratic appointee, seemed to sum up the case when she quipped that the IEEPA “has a lot of verbs … It just doesn’t have the one you want”.

In short, whether such an ambiguous phrase could confer such sweeping powers was sharply questioned by justices on both sides of politics.

Discussion of refunds on tariffs already paid

The fact the Supreme Court went on to consider the question of remedies for potentially striking down the tariffs is also a telling sign.

Specifically, Justice Barrett asked how the process for issuing refunds for the potentially illegally collected tariffs would work.

Counsel for the plaintiffs explained the five businesses that brought the action against Trump’s tariffs would be reimbursed first.

As to the imports from the rest of the world, given the case was not a class action, the process would be “a very complicated thing”. As the lawyers for the businesses elaborated on what the refund process might look like, Justice Barrett interjected with the summation: “So, a mess”.

Counsel for the businesses noted there may be legal precedent for the court to limit its decision to “prospective relief”. This means the Supreme Court’s decision would only affect tariffs collected after the court’s judgement, with no effect on tariffs collected before it.

If this legal precedent were to be followed, refunds would not be issued for tariffs collected before the Supreme Court decision (except for the five businesses that brought the case). The court did not pass any comment on the likelihood of following such a precedent.

Regardless of how the refunds might be issued, it is clear they would result in economic and political upheaval, both for the US and exporters from around the world.

Nonetheless, counsel for the businesses noted the Supreme Court had previously said in a case from 1990, “a serious economic dislocation” was not a reason not to do something. In other words, the fact the reimbursement process would be difficult to administer should not be a block to the Supreme Court ruling the tariffs are illegal.

When will the justices rule?

The court agreed to hear the case on an “expedited” basis, but has not set a date for when it will rule. Betting markets were swift to react, though, with traders marking down the chances of the court ruling in Trump’s favour to 30% after the hearing, from nearly 50% before.

Never one for understatement, Trump has said, “I think it’s the most important decision … in the history of our country”.

Despite Trump’s hyperbole, the case currently before the US Supreme Court is not just about the “liberation day” tariffs. It is also about the role of the judiciary in limiting ever-expanding presidential power. This role is so important that it transcends political lines.

Catherine Gascoigne, Macquarie Research Fellow in International Economic Law, Macquarie University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Kieran Hegarty, RMIT University; Anthony McCosker, Swinburne University of Technology; Jenny Kennedy, RMIT University; Julian Thomas, RMIT University, and Sharon Parkinson, Swinburne University of Technology

In the short time since OpenAI launched ChatGPT in November 2022, generative artificial intelligence (AI) products have become increasingly ubiquitous and advanced.

These machines aren’t limited to text – they can now generate photos, videos and audio in a way that’s blurring the line between what’s real and what’s not. They’ve also been woven into tools and services many people already use, such as Google Search.

But who is – and isn’t – using this technology in Australia?

Our national survey, released today, provides some answers. The data is the first of its kind. It shows that while almost half of Australians have used generative AI, uptake is uneven across the country. This raises the risk of a new “AI divide” which threatens to deepen existing social and economic inequalities.

A growing divide

The “digital divide” refers to the gap between people or groups who have access to, can afford and make effective use of digital technologies and the internet, and those who cannot. These divides can compound other inequalities, cutting people off from vital services and opportunities.

Because these gaps shape how people engage with new tools, there’s a risk the same patterns will emerge around AI adoption and use.

Concerns about an AI divide – raised by bodies such as the United Nations – are no longer speculative.

International evidence is starting to illustrate a divide in capabilities between and within countries, and across industries.

Who we heard from

Every two years, we use the Australian Internet Usage Survey to find out who uses the internet in Australia, what benefits they get from it, and what barriers exist to using it effectively.

We use these data to develop the Australian Digital Inclusion Index – a long-standing measure of digital inclusion in Australia.

In 2024, more than 5,500 adults across all Australian states and territories responded to questions about whether and how they are using generative AI. This includes a large national sample of First Nations communities, people living in remote and regional locations and those who have never used the internet before.

Other surveys have tracked attitudes towards AI and its use.

But our study is different: it embeds questions about generative AI use inside a long-standing, nationally representative study of digital inclusion that already measures access, affordability and digital ability. These are the core ingredients people need to benefit from being online.

We’re not just asking “who’s trying AI?”. We’re also connecting the use of the technology to the broader conditions that enable or constrain people’s digital lives.

Importantly, unlike other studies of AI use in Australia collected via online surveys, our sample also includes people who don’t use the internet, or who may face barriers to filling out a survey online.

Australia’s AI divide is already taking shape

We found 45.6% of Australians have recently used a generative AI tool. This is slightly higher than rates of use identified in a 2024 Australian study (39%). Looking internationally, it is also slightly higher than usage by adults in the United Kingdom (41%), as identified in a 2024 study by the country’s media regulator.

Among Australian users, text generation is common (82.6%), followed by image generation (41.5%) and code generation (19.9%). But usage isn’t uniform across the population.

For example, younger Australians are more likely to use the technology than their elders. More than two-thirds (69.1%) of 18- to 34-year-olds recently used one of the many available generative AI tools, compared with less than 1 in 6 (15.5%) 65- to 74-year-olds.

Students are also heavy users (78.9%). People with a bachelor’s degree (62.2%) are much more likely to use the technology than those who did not complete high school (20.6%). Those who left school in Year 10 (4.2%) are among the lowest users.

Professionals (67.9%) and managers (52.2%) are also far more likely to use these tools than machinery operators (26.7%) or labourers (31.8%). This suggests use is strongly linked to occupational roles and work contexts.

Among the people who use AI, only 8.6% engage with a chatbot to seek connection. But this figure rises with remoteness. Generative AI users in remote areas are more than twice as likely (19%) as metropolitan users (7.7%) to use AI chatbots for conversation.

Some 13.6% of users are paying for premium or subscription generative AI tools, with 18 to 34-year-olds most likely to pay (17.5%), followed by 45 to 54-year-olds (13.3%).

Also, people who speak a language other than English at home report significantly higher use (58.1%) than English-only speakers (40.5%). This may be associated with improvements in the capabilities of these tools for translation or accessing information in multiple languages.

Bridging the divide

This emerging AI divide presents several risks if it calcifies, including disparities in learning and work, and increased exposure for certain people to scams and misinformation.

There are also risks stemming from overreliance on AI for important decisions, and navigating harms related to persuasive AI companions.

The biggest challenge will be how to support AI literacy and skills across all groups. This isn’t just about job readiness or productivity. People with lower digital literacy and skills may miss out on AI’s benefits and face a higher risk of being misled by deepfakes and AI-powered scams.

These developments can easily dent the confidence of people with lower levels of digital literacy and skills. Concern about harms can see people with limited confidence further withdraw from AI use, restricting their access to important services and opportunities.

Monitoring these patterns over time and responding with practical support will help ensure the benefits of AI are shared widely – not only by the most connected and confident.

Kieran Hegarty, Research Fellow, ARC Centre of Excellence for Automated Decision-Making & Society, RMIT University; Anthony McCosker, Professor of Media and Communication, Director, Social Innovation Research Institute, Swinburne University of Technology; Jenny Kennedy, Associate Professor, Media and Communications, RMIT University; Julian Thomas, Distinguished Professor of Media and Communications; Director, ARC Centre of Excellence for Automated Decision-Making and Society, RMIT University, and Sharon Parkinson, Senior Research Fellow, Centre for Urban Transitions, Swinburne University of Technology

This article is republished from The Conversation under a Creative Commons license. Read the original article.