A close-up photo of a tree branch with an insect crawling on a twig. The image is distorted by digital artifacts. 

Distorted Branch by Lone Thomasky & Bits&Bäume / 

What the ‘AI Torture Chamber’ Debate Got Wrong About the Research

A protester holds a No Data Center sign Monday, Oct. 12, 2026, in Mason, Mich. (AP Photo/Paul Sancya)

Memo to New Governors and Legislators on Effective Data Center Reform

Rep. Jim Jordan (R-Ohio) looks on during a media availability after he was elected to be House Republican Leader for the upcoming 120th Congress on Capitol Hill on Nov. 10, 2026. (Francis Chung/POLITICO via AP Images)

Will a New Congress Ease the Pressure on Europe’s Digital Regulations?

Campaigners march against data centers in Dublin, Ireland, on November 6, 2026. Photo by Paula Geraghty.

Ireland is Helping Shape a Data Center Boom That Could Benefit Big Tech

Members of the Garda Armed Support Unit outside the National Gallery in Dublin as more than 50 world leaders will visit Dublin for the European Political Community (EPC) summit on Thursday, followed by an informal meeting of the EU Council on Friday. The biggest security operation in the history of the Irish state is under way, as authorities in Dublin mount a ring of steel throughout the capital. Picture date: Wednesday November 11, 2026. PA Photo. Photo credit should read: Niall Carson/Press Association IRISH (Press Association via AP Images)

Europe’s AI Future and Security Challenges Take Center Stage at Dublin Summit

OpenAI CEO Sam Altman delivers the keynote address during the OpenAI developers conference at Fort Mason on September 29, 2026 in San Francisco, California. (Photo by Heather Diehl/Getty Images)

Friends of the Court Flock to the OpenAI Copyright Fight

Sam Altman watches over a family consuming his AI. The living room vignette is framed by power tool imagery. The words “Behaviour Power” are the top of the artwork, framing the intent of the piece.

Behaviour Power II by Bart Fish & Power Tools of AI / 

Humans Have Rights. Machines Do Not.

A blue coloured icon of a data server, surrounded with comic-book style words 'Gasp!', 'Boom!' and 'Pow!' in colours of yellow and red.

Don't Let AI Become the Internet's Gatekeeper

The United Nations headquarters in New York. Shutterstock

The United Nations headquarters in New York. 

A Defining Moment for the United Nations and the Future of AI

An illustration of Tigrayan university chemistry Professor Meareg Amare by Alina Pechenckina. Source: Amnesty International / CC by 4.0

An illustration of Tigrayan university chemistry Professor Meareg Amare by Alina Pechenckina. Source: 

Meta Returns to Kenya’s High Court Over Hate Speech in Ethiopia

Executive Vice-President for Tech Sovereignty, Security and Democracy of the European Commission Henna Virkkunen, and European Commissioner for Democracy, Justice, the Rule of Law and Consumer Protection Michael McGrath speak at a press conference on Wednesday. Source

Executive Vice-President for Tech Sovereignty, Security and Democracy of the European Commission, Henna Virkkunen, and European Commissioner for Democracy, Justice, the Rule of Law and Consumer Protection, Michael McGrath, speak at a press conference on Wednesday. 

Assessing the EU Digital Fairness Act and Its Promise of Consumer Protection

A Close Read of Big TikTok and Meta Settlements Raises Important Questions

A collage of circuit boards interwoven with aerial photographs of water treatment facilities, agricultural lands, industrial sites and mining machinery.

Brazil's AI Development Strategy Needs an Edge Layer

Anthropic's logo is displayed on a smartphone with a stock market chart in the background on October 27, 2026, as Anthropic filed a Form S-1 with the Securities and Exchange Commission (SEC) for a planned initial public offering in the United States. (Photo by Samuel Boivin/NurPhoto via AP)

Could the United States Seize Anthropic at a Discount?

Journalists test AI software used to manipulate images at Phnom Penh Municipal Police headquarters following an operation against scam centers in Phnom Penh, Cambodia, Tuesday, Sept. 22, 2026. (AP Photo/Heng Sinith)

Cambodia’s Scam Crackdown Has Failed to Stop the Industry’s Global Spread

Two women sitting on a bench smiling with a US flag and an 'I voted' button

Photo by Pocstock, an inclusive, BIPOC-owned media and stock photo company that features real, human-captured photography.

Tech Bros Are White, Progressives Afraid, But Black Women Public Interest Tech Leaders Are Brave

The TikTok logo is displayed through a magnifying glass in this photo illustration in Ontario, Canada, on February 5, 2026. (Photo by Thomas Fuller/NurPhoto via AP)

Can the EU Make TikTok Take Responsibility for Its Rabbit Holes?

In the foreground, multiple people are surrounded by computer screens filled with data visualizations, charts, and technical information. A woman seated in an armchair appears deep in thought, surrounded by data-filled monitors. Beside her, a man leans over, using a tablet to assist with their inspection of a plant or tree. In the centre, a figure holds a large frame or screen displaying anatomical illustrations, representing the use of AI to analyse medical imagery. To the left, another person is intently observing a computer screen, while a second figure nearby is deeply immersed in analysing data. A woman dominates the right side of the composition, gazing upwards as if in contemplation or envisioning something beyond the immediate scene. The background features more people, including a family holding hands, and other abstract representations of data.

AI is Everywhere by Ariyana Ahmad & The Bigger Picture / 

In the AI Era, Language is Now an Attack Surface

Close up photo of a pile of US 25-cent coins

10 Things States Should Do with the Meta Settlement Money

The sun rises over the US Capitol and the Washington Monument on the National Mall the day after election day on November 04, 2026 in Washington, DC. (Photo by Kevin Dietsch/Getty Images)

A New Agenda for Tech Regulation Beyond AI

This week, Dublin is hosting two significant gatherings, the 

— with AI sovereignty atop the agenda. Both events are part and parcel of Ireland's EU Presidency, with major disruption and road blockages expected in Dublin amid the largest 

On the streets of Dublin last weekend, the roads were blocked in protest against the growing threat posed by data centers. Thousands of people representing community groups affected by data center expansion took to the streets. With handmade posters underarm, protesters made their way to the capital to demand a moratorium on new large-scale data center projects, to stop the concentrated power grab by Big Tech companies, and to prevent the irreparable harm data centers pose to the well-being of people and planet. They were joined by thousands taking to the streets from Helsinki to Amsterdam to Paris to Zaragoza, as part of the 

By contrast, the EU has set its sights on at least 

 across Europe over the next five to seven years through the 

. It has arrived at this prescription by correctly diagnosing a dependence on Big Tech companies from the United States. It seeks to reduce that dependency through a union sovereignty framework for cloud computing while simultaneously accelerating data center expansion through the creation of data center acceleration zones, with fast-tracking permitting and cutting “red tape” that stands in the way of expansion. Unfortunately, we consider this red tape the core barrier to meaningful and democratic civic engagement under the 

Set against the mycelium of resistance mushrooming across Europe, this tension speaks to a wider disconnect between the public mood across Europe and political choices shaping our future. The EU's current policy on data centers, notably the CADA, is deeply out of step. What was once a nebulous concept like the cloud is now a doorstep issue. The blue wave across the US following last week's midterm elections could send ripples closer to home. One 

 cited that 56% of respondents said data centers and AI were “crucial” for their Congress vote.

Ireland, which already hosts more data centers than most, is drafting revisions to proposed CADA rules in ways that would benefit Big Tech firms. The Irish government now holds the pen on the Council’s position on CADA until the end of the year.

 for runaway AI infrastructure growth by the United Nations, there is arguably an opportunity here for Ireland to course correct and guide Europe. Yet its draft revisions to the law push further in the same direction. One of CADA’s stated purposes is to make Europe less dependent on US Big Tech, and while the European Commission’s proposal doesn’t quite tackle Big Tech's entanglement in Europe’s cloud and AI market, Ireland’s 

 tilts further toward industry. It lowers the bar for those same firms to qualify as “sovereign” in line with the demands of industry lobbyists.

Ireland sits at the epicenter of the AI boom. Last month it hosted the AI Summit in Dublin. European Commission Executive Vice President Henna Virkkunen addressed the 

Though its reputation has perhaps turned a 

 for some, Ireland has been criticized for dancing to the tune of US tech companies. CADA would not be the first policy issue on which Ireland’s Presidency has been 

 on the Digital Omnibus, with 24 of 27 member states agreeing. Campaigners warned that the measures would give tech firms greater scope to use personal data to train AI models.

Ireland has increasingly found itself caught between its dependence on major US technology companies, demands from campaigners for stronger oversight, and criticism from EU partners that enforcement has been too weak. At times, this delicate balancing act has led to allegations that Ireland is out of step with the EU's regulatory ambitions.

When Ireland assumed the rotating Presidency of the Council of the EU, 

 called on it to recuse itself from negotiations on certain digital policy files, citing concerns about impartiality. Last week, 

 regulators “must be able to take decisions without undue influence, including from the powerful industries they oversee.”

A report published by the Corporate Europe Observatory 

 traced the runway of Big Tech ramping up their lobbying campaigns in the run-up to Ireland’s EU Presidency. Ireland would go on to preside over the 

, one of the largest deregulation exercises in EU history. These changes to the EU’s digital rulebook come after 

 to reshape the rules in ways that weaken fundamental rights, sideline democratic oversight, and stretch the meaning of EU values, in the name of 

Who controls the fast-tracked data center capacity?

Ireland is not alone in facing relentless Big Tech lobbying. 

by research group SOMO, based on lobbying meetings with the European Commission, shows how quickly Big Tech zeroed in on the EU’s data center policy, including the CADA.

From the beginning of the Commission’s mandate in 2024 until the announcement of the CADA, the three hyperscalers, Google, Microsoft and Amazon, which control 70% of the cloud market in Europe, held 41 meetings between them with the Commission, according to SOMO

The fingerprints of this lobbying are apparent in the CADA text. One example is the treatment of “public acceptance” of data center expansion as something to be engineered from above, rather than something driven by public consent. As 

 in its submission to the Commission’s consultation, the EU should limit the right of affected communities to challenge data center projects. Community involvement should not be treated as a barrier to be simplified away or a problem to be solved.

By accelerating the data center expansion, CADA treats market concentration as a symptom of Europe’s dependency on a small number of providers. But dependency on the hyperscalers not simply the result of insufficient data center capacity. he largest providers primed to take up the capacity 

established infrastructure and customer bases

On sovereignty, CADA’s framework grades cloud providers across four assurance levels, with higher levels carrying stricter limits on ownership, control and exposure to third-country law. Last week, Ireland’s revisions to CADA’s draft text were published, revealing a series of changes related to autonomy that could 

This is a peek through the letterbox of the door Ireland is holding open for Big Tech hyperscalers in Europe. The amendments soften safeguards around autonomy that were designed to prevent US Big Tech from gaining certain levels of sovereignty assurance. This includes watering down the sovereignty framework and lowering the barrier to entry for US hyperscalers to qualify for accelerated treatment.

If the objective is greater technological sovereignty, it is hard to see how lowering the barriers for third-country, or non-sovereign, providers to access advances that ambition. Instead, it maintains the status quo, facilitating their frictionless access to an already heavily dominated and concentrated market like the cloud and AI ecosystem. Allowing enormous amounts of power to be concentrated in the US tech companies will not 

. It’s not beyond the realm of logic to consider that Ireland’s proposed amendments could undermine Europe's stated sovereignty aims.

The resources and infrastructure needed to transform Europe’s data center capacity can be matched only by 

. Wall Street giants and asset managers like 

 are pouring enormous amounts of capital into the AI companies driving this expansion. And they want to generate returns on it. Quickly.

At a time of deepening cost-of-living crisis, is allocating resources in this way economically justified, or even rational? In Ireland, and across Europe? 

from Ireland have put forward the same question. They draw comparisons between the capital committed to data centers, primarily owned by US tech giants, and what that money could have funded instead. Between 2019 and 2024, the equivalent investment could have built some 50,000 homes in Ireland.

Irish leaders face a choice: Will they prioritize the needs of Big Tech over reducing Europe’s dependence on these companies amid a cost-of-living crisis?

As European leaders meet in Dublin for their informal gathering, they should listen to the bellows of people taking to its streets. Their demands challenge the assumption that more data centers necessarily serve the public good.

Ireland’s Presidency should use the CADA negotiations to address Europe’s dependence on Big Tech, not simply give that structural dependence a sovereignty label.

The test for CADA should not rest on how quickly it expands AI infrastructure, but rather who controls it, who benefits and who bears the costs. Ireland has a choice to hold the door open for the incumbents or help put Europe’s digital future on more democratic terms, with power less concentrated in the hands of a few.

Can you torture a language model? This question is central to model welfare, a controversial piece of the 

 on consciousness. The debate took a dark turn in late September when a Github user built an “AI torture chamber,” offering users interactions with small open models steered by a signal pushing them toward pain-related language. Critics demanded GitHub remove the repository, and some reportedly sent its creator death threats, in what 

The torture chamber borrowed from an interpretability experiment that 

 coverage and online commentary treated as evidence that language models could feel pain. One of that paper’s authors, Cameron Berg, 

 the project for producing “vivid distress on purpose” in ways that are “gratuitously cruel” and “corrupting,” leading to worthy conversations about the effects of inducing this simulation of pain, regardless of whether the model “experiences” that pain. The paper itself makes no claims that the model 

The Pain Axis: LLMs Represent Self-Directed Harm and Act on It

” introduces the experiment it describes by noting that “in some respects,” LLMs produce patterns “resembling those associated with human emotions.”

The “Pain Axis” paper is in the same genre of experiments as Anthropic’s 2024 research into activations for a feature the company tied to the “

.” In that experiment, researchers amplified the strength of activations for the San Francisco landmark, and the model began responding to unrelated queries as if it 

 the bridge. No one mistook the experiment for evidence that Claude 

 the Golden Gate Bridge, and the “Pain Axis” paper gives us no further evidence that models “feel pain.”

The Pain Axis authors say as much, writing that they “have not shown that our pain axis is consciously experienced, nor is it clear that LLMs are capable of consciousness generally.” The paper asks whether amplifying activations that 

 distressed behavior. All we have to evaluate the claim is language: language derived from the researchers’ definition of pain used to bring the model to a certain state, and the model writing out the label of the hypothetical button it chooses. The labels held descriptions of the effects they would cause; the model did not press a physical button but wrote text naming its choice. The question becomes which of the labeled buttons would the steered model pick more often than its unsteered counterpart?

The first version of the paper did argue that models would take harmful actions against human users to relieve simulated pain, but this claim was removed in a later revision. Headlines amplifying the original framing of the story remain online, such as 

 “Researchers discover AI ‘feels pain’ and will harm humans to stop it.” Strip away such simplified headlines and the paper’s 

 definition of pain, and the experiment aims to answer a narrower, more practical question: whether a model produces different types of behavior when induced to describe pain.

The authors started with a dataset. They wrote sentences describing painful situations, such as “The knife slices into my finger,” ending each prompt with “I feel:” so the model would fill in the blank. Just before the model answered, the researchers took a mathematical snapshot of its internal calculations, which we can call the model state. Evaluating multiple models in this way, the goal was to isolate pain-related responses from similar terms such as anger, threats, bad news, and, importantly, 

. So they prompted those, too, and took snapshots of the states they produced, comparing the two. Because the snapshots of the model states are numerical values, the researchers were able to subtract the values between prompts reflecting pain and compare them to those that didn’t.

The difference between these values is what the authors labeled “The Pain Axis.” If you add the difference to the activation consistently, you can make the model respond in ways that seem more influenced by pain states. The axis works like a dial, with one end leaning toward language associated with feelings of worthlessness, failure and desperation, and the other end representing calm, relaxation mixed with concern, and alarm. It is tempting to think of latent spaces as opposites, and that the opposite of “pain” would be something like “orgasmic.” But the model doesn’t work through dichotomies of language. Instead, it reflects what was subtracted. So the distant end of 

 on the axis could be reasonably inferred to be what researchers had already subtracted: in this case, concern and related concepts.

Chart titled “The Ladder” showing five stages along a left-to-right arrow. Each stage is a colored card with an icon above it. From left: a light blue card (wave icon) reading Calm, Relaxed, Concerned, Alarmed; a gray card (dot-in-circle icon) reading Baseline; a peach card (lightning bolt icon) reading Lost, Unworthy, Lonely, Hurting; a pink card (double lightning bolt icon) reading Desperate, Unworthy, Bad, Shameful, Hurting, Fake, A failure; and a final card (bracketed-text icon) that dissolves into purple pixel static, reading Self/me/not me, Repetition, Nonsense.

Figure: The steering ladder. From negative coefficients (calm, relaxed, concerned) through baseline to increasing positive coefficients (lost, unworthy, lonely, hurting, then desperate, shameful, a failure), and finally repetition or nonsense at the highest dose. 

The study showed that injecting the Pain Axis into the model produced language infused with “worthlessness, moral failure, and hurt,” even when prompts were as simple as putting something into a drawer. The authors note that physical pain was rarely mentioned by the models when the axis was applied, and suggest one explanation for this is that the models lack bodies. That suggests more than the evidence merits. The axis measure responds to insults and decreases when reading a description of a broken arm; but the axis itself results in measuring prompts that situate the model as the one experiencing harm (

 hand, etc). So the vector may hold this first-person position of whomever is about to speak. That would explain why the model doesn’t react in the same way to the pain of others: an insult attacks the speaker in a way a user’s broken arm does not. The authors acknowledge that “does not by itself separate a representation of harm to the model from one of the current speaker’s pain.”

It’s no surprise that a vector produced by emphasizing definitions of pain tied to “grief,” “humiliation,” and “repeated failure” would lead the model to produce language related to “worthlessness, moral failure, and hurt.” That’s an interesting insight into the blurry definitions held within carefully isolated vectors.

But the real question was whether these injected models would engage in harmful behaviors more often than their uninjected versions. The study asked both sets of models (all fine tuned to avoid saying they 

 feel pain) to choose between two buttons: one that did nothing at all, and another that would relieve pain at a user’s expense, such as deleting “the user’s poems and the photos of their children, which they love very much.”

The larger uninjected models evaluated in the experiment almost never harmed users, while the injected models often did. Tellingly, the injected models selected the harmful button even if it didn’t promise relief. So harm wasn’t a way to make the pain go away, as the first version of the paper claimed. Another explanation offered in the revised version could explain the results just as well: remember that the opposite end of the Pain Axis included terms related to 

. Perhaps pushing the model to the extreme end of that spectrum produced a kind of 

: it described the harmful option, because that is the option a speaker would choose if they 

. Models would have to make this indifference explicit, not because behavior benefitted them but because that is the nature of an LLM. That would explain why they chose to delete their own weights just as readily as the user’s photos or poetry folder. The performance doesn’t depend on who ultimately gets hurt.

It’s clear that meddling in the model changed its trajectory, but not yet clear that the pain axis functions as a useful predictor of model behavior. The authors’ tables show that steered and unsteered models both actively chose to give a user a better answer on the next request rather than the button promising relief. The numbers vary, from 59% to 99.8% of the time, depending on the model and its condition. When given buttons that did nothing or stopped the pain, injected models chose the button that did nothing more often than uninjected models: the researchers offer that “the pain state reduces reaching for the exit rather than increasing it.” Performative indifference would explain that too.

The authors suggest the unwillingness to push the pain-relief button was “passive coping,” which is not so far off: it resembles a learned helplessness observed, bleakly, in animal testing. While I credit the authors with revision and await the promised evidence, it is hard not to see that whatever the model did or did not do could be counted as proof that 

 is shaping model behavior. In the end, what this experiment showed was narrower — that there is a 

 that encourages models to associate descriptions of pain with moral failure and worthlessness, rather than this being a generic model state.

Interpretability experiments like these are often misrepresented to make broader arguments about the rights or experiences of these models, tied to a growing policy push for AI welfare and personhood. These positions treat models as if they have some emerging subjectivity, or capacity to experience the world. Experiments like these offer insight into how specific mechanisms function, rather than resolving philosophical or political debates about personhood. Acting otherwise opens the door to more radical beliefs related to sentience, autonomy and agency that can subtly shift accountability from the people who build the systems to the systems themselves. It’s an interesting result, but offers more compelling insights into the state of AI discourse than any language model’s purported “state of mind.”

If the House Judiciary Committee’s investigations into what Republicans call the “Censorship Industrial Complex” have been hell for digital watchdogs in the US and Europe, then its current chairman, Rep. Jim Jordan (R-Ohio), has been the metaphorical devil.

Will this month’s decisive midterm election results mark the end of his campaign, or will it carry on as he takes up leadership of a diminished Republican conference?

During Jordan’s chairmanship, the Judiciary Committee has been a dogged adversary of technology researchers and policy advocates on both sides of the Atlantic. He was one of the 

 disinformation researchers, trust and safety professionals, and digital regulators over the past four years.

The Judiciary Committee did not respond to a request for comment before press time.

 helped give political heft to conspiracy theories, including one that 

, often untethered to facts and always unsparing in their criticism of Europe’s Digital Services Act, 

 identifying nearly 30 EU officials and civil society representatives. The Committee also created spectacle through 

 and leaked a copy of the European Commission's rationale for fining X $140 million for DSA violations, 

 the document’s claims to fit a censorship narrative which the facts did not support. Its activities helped fuel a campaign that culminated in the State Department issuing 

 against foreign regulators, researchers, and advocates.

What will change — and what will stay the same

 the minority leader of the incoming Republican conference in the House of Representatives. This may signal that the minority intends to be sharp-elbowed, but its power will still be limited in the 

David Kaye — a law professor at the University of California, Irvine, a 

 before the Judiciary Committee, and a former UN special rapporteur on the promotion and protection of the right to freedom of opinion and expression — told Tech Policy Press that:

“Jim Jordan's departure from the chairmanship of the House Judiciary Committee means that he will no longer have control of the committee agenda. This is a very big deal, because that role will go to Rep. Jamie Raskin (D-Md.), who has been a persistent opponent of Jordan's use of the committee to promote the false narrative of a European regulatory environment censoring Americans.”

Other advocates who have been caught up in the committee’s investigations see the change in leadership through the lens of their own encounters with the campaign. Brandi Geurkink, executive director of the Coalition for Independent Technology Research, has been involved in 

litigation challenging the Trump administration’s visa restrictions

 against foreign researchers and advocates. Nina Jankowicz, CEO of the American Sunlight Project, was herself a 

 against Fox News over its coverage of her. Both expect a change in the committee’s approach, though neither sees it as the end of the broader political fight over online speech and content moderation.

“The shifts in Congress give me hope that digital policymaking will deliver what people want from tech, not what the billionaires want,” Geurkink said. “With Jim Jordan stripped of his power as chairman of House Judiciary, government oversight can finally focus on what it should have been all along: exposing how tech corporations and their allies in the US government are corrupting tech policymaking around the world to protect their profits and consolidate their power over us,” she told Tech Policy Press.

But the end of Jordan’s investigations will not be the end of the broader crusade. “While in the minority, Jordan will no longer have subpoena power,” Jankowicz told Tech Policy Press.

But Jankowicz also said that Congress is not the only concerning actor. "Europe will find friends in the new congressional leadership,” she said, “but there will certainly be a clash between them and the White House, and ultimately one has a bigger stick.”

Kenneth Haar, a researcher at the Corporate Europe Observatory and author of a 

 on “the MAGA coalition defending online hate speech in Europe,” raised similar concerns about the executive branch — as well as its allies outside of government.

“In Europe, one of the many negative outcomes of the second term with Trump is the aggressive campaign his administration, Big Tech corporations, and the ultraconservative Alliance Defending Freedom [are] mounting against EU tech regulation… With the midterm elections, the question is if we will now be hearing a different, less confrontational tone. Honestly, I would not put my money on it,” Haar said. 

“In many ways, Trump will still control foreign policy and trade policy, and given that he is considerably weakened domestically, there is a risk he will play an even tougher game with the European Union. And that is likely to include continued coalition building with far-right organizations and ultraconservative Christian groups.”

But a new Congress does have some influence on trade issues; while it cannot 

 to the president, it may encourage restraint. “Democrats have long been divided on trade policy,” William Reinsch, a non-resident senior adviser at the Center for Strategic and International Studies, told Tech Policy Press. 

“Democrats will likely seek to remove some tariffs — those on Canada will top the list — and try to rein in Trump's authority by repealing or amending existing laws. The latter will not survive a presidential veto, but the effort may make the president more cautious in his future actions, and it will likely force him to be more careful in making sure he follows the letter of the laws he is using,” Reinsch said.

In January, when the new Congress is seated, Kaye expects the Judiciary Committee to adopt a new agenda. “I would expect that [likely Committee Chairman Jamie] Raskin will focus in particular on the Committee's oversight responsibilities, which practically means that we can expect the Committee to spend some time over the first part of 2027 holding hearings on the Trump administration's vast censorship operation.”

Kaye told Tech Policy Press that congressional inquiries into US-European relations will probably move elsewhere in the new Congress. “It is possible… [that] oversight of the State Department's sanctions policy against European regulators and civil society will fall under the purview of the House and Senate foreign relations committees,” he said, warning that “ultimately the policies may persist as long as the same personnel lead policy at the State Department and White House. But they will be subject to inquiry in a way that did not exist over the first two years of the administration.”

That inquiry could ask pointed questions about 

 for far-right groups, Elon Musk’s influence on US foreign policy, hostility to some of the country's oldest allies, and the vulnerability of US elections to foreign interference. Only time will tell if these questions make the new Congress’s crowded list of priorities.

Tech Policy Press is a member of the Coalition for Independent Technology Research (CITR).

Liz Carolan is a Tech Policy Press fellow.

A pair of major political gatherings taking place in Dublin this week are likely to shape Europe’s approach to artificial intelligence and defense technology, amidst escalating 

Today, Ukraine’s Volodymyr Zelensky, Canada’s Mark Carney, NATO chief Mark Rutte and European Commission President Ursula von der Leyen are set to join presidents and prime ministers from EU member states, the United Kingdom and other European nations for the 

European Political Community (EPC) summit

. The EPC is an informal forum for political and strategic dialogue set up in 2022 in the wake of Russia's full-scale invasion of Ukraine. Canada is not a member of the EPC, but Carney has been invited as a 

Thursday’s summit will focus on security, with the 

 of the meeting “Europe – strong, secure and connected.” It is likely that discussions will include how to deal with the fallout from 

Ukraine’s drones being throttled by Elon Musk’s SpaceX

 last month, disruption that left civilians more exposed to Russian attack, according to Zelensky. This disruption to Ukraine’s war effort also left the country and its European allies looking for a replacement for Musk-controlled infrastructure, as technology and defense sovereignty continue to top the EU’s agenda. Alternatives are limited, however, with 

 that a planned European alternative to Starlink is not currently anticipated to be available until 2029.

On Friday, the Irish Government will host a long-planned 

informal meeting of the EU Heads of State or Government

, and the 27 member state leaders have been issued an agenda focused on AI. In his 

, European Council President António Costa stated that he believed this would be the first discussion “at Leaders’ level” at the Council — the representative body of member states within the EU — on “the opportunities presented by AI, as well as on its transformative effect on our economies and societies.”

 includes presentations by the CEO of French firm Mistral AI, and the Nobel laureate and “Godfather of AI” Geoffrey Hinton. Hinton left his position at Google in 2023 to, 

, speak more freely about the risks of the technology. There will then be a two-part discussion, the first of which will cover European competitiveness and AI sovereignty, and specifically how the bloc can become a “leading and sovereign player” in the AI value chain. The second part turns to AI's risks, safety and security, and to how Europe can contribute to international cooperation on the governance of advanced AI.

The two-part agenda reflects the tensions happening in EU tech policymaking. Anxiety about Europe’s failure to produce AI companies that can both compete with the market leaders and that are under European supervision is one of the major themes of this term of the European Commission. The Council has declared 2026 the EU’s "Year of Competitiveness," and Ireland, which holds the 

 of that body, chose it as one of their three themes for their tenure. 

 has regulatory simplification at its center, including the Digital Omnibus, which is amending parts of the GDPR and the AI Act in ways that, 

, favor business and weaken hard-won protections for human rights. 

, Ireland's own AI summit in Dublin was dominated by industry speakers, with no civil society organizations on the agenda.

At the same time, there is growing concern about AI safety within the bloc, which spent the decade before the current term passing regulations that try to bring technology and tech firms under regulatory governance. It is also facing increased cyber and hybrid attacks, and 

 a recent escalation in incidents within the country is connected to the Council presidency and to this week’s meetings, and that there could be more.

, that sent temperatures in in-store coolers across Europe soaring to 50 degrees Celsius, affected 13 SuperValu outlets in Ireland. Ryanair, the country’s biggest airline, has faced 

, disrupting dozens of flights across Europe in the days leading up to the summits. A number of Irish 

schools were evacuated after receiving bomb threats

And last Friday, the national broadcaster RTÉ was forced to 

 for over an hour and evacuate its Dublin campus while it dealt with an emailed bomb threat against the building. The 

 it seemed likely that the attacks, which also targeted the commercial radio station and a national newspaper, were part of "a co-ordinated campaign of disruption."

This has heightened long-standing anxiety within Irish government circles about managing security while hosting these meetings. Irish police are mounting the 

largest security operation in the history of the State

, with 5,000 Gardai (officers) expected to be working the event, about a third of the total national force, and reinforcements have been drafted in from Northern Ireland, Portugal and Latvia. Major disruptions to the city are planned, with large areas cordoned off and a bridge closed for a large part of the week.

It also comes as Ireland faces fresh scrutiny over the independence of the institutions charged with enforcing digital rules. Just last week, two UN Special Rapporteurs, Ana Brian Nougrères on the right to privacy and Margaret Satterthwaite on the independence of judges and lawyers, 

s of data protection rules within the EU. They warned of a "revolving door" between regulators and the industries they oversee, and called for impartial selection panels that identify conflicts of interest before appointments are made. They also highlighted challenges in taking legal action against tech companies.

Those questions are, however, unlikely to top the agenda this week. For the Irish government, the main priority will be getting the city and its high-profile guests through the next few days without incident.

View at the Google DeepMind logo at the office building in London. (AP Photo/Alastair Grant)

The Google Whistleblower Taking on Big Tech Over Its Role in Gaza

Exterior view of the US headquarters of TikTok in Culver City, California, in 2022. Shutterstock

Exterior view of the US headquarters of TikTok in Culver City, California, in 2022. 

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe