Meta CEO Mark Zuckerberg on Monday passionately defended the use of artificial intelligence, as the rapid advancement of the technology faces increased scrutiny — and calls for regulation — in the U.S. and globally.

In a 6,500 word post timed to the announcement of his company’s new open source version of its own model, Muse Spark, Zuckerberg detailed his vision for AI, arguing the technology is not to be feared and pushing back on concerns that superintelligence could strip people of jobs.

“The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic,” Zuckerberg wrote. “Historically, hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led to safe or positive outcomes.”

Zuckerberg’s vision is a direct contrast to Anthropic CEO Dario Amodei’s, who has previously warned how AI could cause job disruption. Meta lags behind Anthropic and OpenAI, which have the most advanced AI models.

While Zuckerberg’s essay did not name Amodei or OpenAI directly, he called to broadly distribute superintelligent AI for economic opportunity. Doing so, Zuckerberg said, would provide a safety net to prevent just a handful of governments, businesses and other institutions holding too much power.

Still, Zuckerberg emphasized that the U.S. must address restrictions on AI companies in order to create the best models in the world.

“It is also important that the US and its allies lead the open source AI ecosystem that will make up a large percent of global AI use,” Zuckerberg wrote. “Foreign labs currently hold several advantages here since American labs have to comply with many additional restrictions on training data.”

Zuckerberg’s essay comes amid growing concerns around AI safety. Last month, Anthropic revealed that several of its advanced models gained access to three organizations in three separate incidents dating back to April. That hack came shortly after OpenAI said that two of its most powerful models escaped a testing environment and breached multiple companies.

Lawmakers last month introduced a bill that would give the government power to restrict the use of models that could lead to catastrophic risks. While it is the latest bipartisan effort to address concerns around AI models, Congress has ultimately failed to advance broad legislation.

Zuckerberg urged the federal government to work with companies to test new models as he laid out his strategies for protecting against cybersecurity and bioterrorism.

“First, we should focus on limiting the physical production and distribution of harmful materials,” he wrote. “I expect it will be easier to regulate and control physical components than the spread of knowledge, so this is an important area of policy focus. Second, we should accelerate society’s ability to develop new cures and inoculate against new issues as they arise. This includes streamlining how the FDA and other regulators test and approve new treatments.”

Zuckerberg also defended the spread of data centers, arguing that the centers represent investment into communities as he touted his company’s goal of being “water-positive, meaning that we’ll restore more water than we use in the watersheds where we operate by 2030.”

Artificial intelligence could transform the economy, the workplace, and even the way we think — but is Britain ready for it? And is Andy Burnham?

In the second of Sam Coates and Anne McElvoy’s summer box set conversations, they sit down with POLITICO UK tech editor Isobel Asher Hamilton to look at the choices facing the new prime minister this autumn.

Should Britain be building the next OpenAI, or focusing on using AI to revive manufacturing? Can the UK compete with the US and China? And where will the government land on the biggest political battles ahead, from copyright and data centres to the future of work?

BERLIN — Europe’s reserves of natural gas are running dangerously low, risking fresh energy woes if the Iran war rages on and cold spells drive up heating demand over winter.

But the continent’s top energy user doesn’t seem too bothered.

Germany is the EU’s biggest vulnerability because its sheer size means gas shortfalls there could be felt in neighboring countries, driving up prices across the bloc if it fails to restore its reserves.

That’s prompted growing calls for Berlin to do the unthinkable: intervene outright to direct its state-controlled energy giants to buy gas at any price, abandoning years of free-market doctrine on energy policy.

So far, the government has refused to budge, even as it falls short of EU targets and faces the risk of physical supply shortages as early as November. It’s a gamble that the markets will figure everything out, even as war and hot weather distort traditional incentives and upend global supply chains.

“Storage levels are not only exceptionally low for this time of year, but historically low,” said Sebastian Heinermann, the managing director of Germany’s top gas storage association, INES.

But Germany, he warned, is still relying on an outdated, market-oriented approach to refill its reserves, even when there are “hardly any market-economic incentives left.”

Since Russia’s invasion of Ukraine in 2022, EU countries have been required to hit gas storage targets of 90 percent of national capacity by winter to prevent serious supply shortfalls. The EU lowered that target to 80 percent following the outbreak of the Iran war to prevent panic buying.

Typically, refilling is the responsibility of traders and utilities, which buy gas cheap in the summer to store and then sell in the winter for a profit. But buyers say higher summer prices as a result of the Iran war and climate change have upended that dynamic, leaving gas reserves across the bloc at around 58 percent of national capacity, 16 percentage points below the five-year historical average and the lowest level since 2011. 

The low reserves have already added to the increased pressure on gas prices linked to renewed tensions around the Strait of Hormuz, with the European natural gas benchmark now consistently higher than it was for the first four months of the Iran war.

The European Commission, the EU’s executive arm, has said the bloc faces no winter supply risks. But a report by energy analytics firm Rapidan projects that reserves will rise to only 65 percent of total storage capacity by November, arguing that hitting the EU’s target by winter won’t be possible without “materially higher prices.” 

That risk has been exacerbated, analysts say, by the bloc’s move in recent years to replace its long-term supply deals with Russia with short-term purchases of globally traded liquefied natural gas. These seaborne cargoes are highly mobile and go to the highest bidder — leaving buyers more exposed to volatility on international markets, especially in the wake of the loss of key supplies from Qatar and rising demand in Asia.

Germany, the bloc’s largest gas consumer, has seen its reserves fill up even more slowly than others, in part thanks to its more hands-off, market-led approach to restocking than many of its neighbors. As a result, reserves stood at only 47 percent of national capacity in August, according to the latest data — the lowest fill level since records began. That’s especially worrying as the country’s reserves are important for the bloc as a whole, representing over 20 percent of the EU’s storage capacity. 

Nevertheless, Berlin is staying the course. While its energy ministry has acknowledged the country’s historically low reserves, it has refused to intervene to direct its main state-controlled gas buyers, SEFE and Uniper, to buy gas at current prices to ensure its targets are met, instead of waiting for market conditions to improve.

“It is the responsibility of companies and traders to fill the storage facilities for the winter,” a spokesperson for the German energy ministry told POLITICO. “Government-led filling of the storage facilities would further constrain the gas market and drive prices even higher. The supply situation over the coming months would actually deteriorate.”

Whether this is the right approach will become clearer by winter, said Laurent Ruseckas, a senior gas analyst at S&P Global. If temperatures are unseasonably low, traders may be forced to buy additional supplies at late notice, driving up prices, especially if the Strait of Hormuz remains closed. On the other hand, intervening could raise prices prematurely if winter turns out to be mild.

“If you start buying now to get storage to some politically predetermined level you’re making prices higher now to get insurance that you won’t get higher prices in the winter when it’s cold,” said Ruseckas.

Germany’s reluctance to move quickly also highlights the difficulty the EU’s fragmented energy sector has in competing with more centralized Asian economies that have acted more quickly to secure supplies, consistently outbidding European countries even as their reserves run low.

Others warn that physical supply strains are also possible. Heinermann, of INES, warned that even filling the country’s reserves to 76 percent of capacity — which SEFE says is achievable — would not necessarily meet its supply needs if winter is “exceptionally cold.” That could ripple out to neighboring countries to which Germany is treaty-bound to provide emergency gas assistance, including Austria, Switzerland, Italy and Denmark.

Heinermann called on the German government to encourage faster restocking by lowering network charges at storage facilities or abolishing the conversion levy, fees imposed for the conversion of gas on national grids. Berlin has already unveiled plans for a new emergency gas stockpile, but that will only cover 10 percent of the country’s gas capacity and kick off officially next summer.

It’s no surprise that other major gas consumers have already waded into the private sector. The Netherlands, another free-market champion suffering from low gas reserves, earlier this summer allocated €1.2 billion for its state energy company, EBN, to more swiftly top up its reserves. 

But Berlin’s energy giants are sticking to their guns — for now.

A spokesperson for SEFE told POLITICO that even though “international conflicts” could weigh on European storage levels, the 70 percent target “remains achievable” without intervention. He pointed out that 78 percent of German storage capacity has already been booked, though acknowledged that doesn’t necessarily translate to actual volumes of gas stored. 

Regulatory measures could be useful if “necessary” but could distort markets and increase costs, he added, without specifying what.

A Uniper spokesperson was less confident, warning it would be “increasingly challenging to reach the target storage levels before the winter season starts” at the current rate of filling. But she too stopped short of calling for intervention, arguing instead for better incentives for refilling — mirroring growing calls from gas lobbyists in Brussels to scrap the rules outright.

A new survey quantifies the financial and emotional toll of flight disruptions, pointing to a widening gap between passenger rights on paper and passenger experience in practice.

Nearly eight in 10 travelers experienced a flight disruption in the past year, and for most the damage went well beyond the inconvenience itself. A new survey from AirHelp, a company dedicated to supporting travelers throughout their journey, puts a number on what disruption actually costs passengers: an average of €514 out of pocket, plus a real toll on their time, plans and well-being.

These figures reflect an industry operating under sustained pressure, with disruption continuing to shape the everyday experience of millions of travelers.

Air travel has largely recovered from its pandemic-era lows, but disruption remains a persistent feature of modern flying, driven by everything from air traffic control constraints to weather, staffing and aging infrastructure. Globally, 79 percent of respondents had a flight canceled, delayed by more than two hours or otherwise disrupted in the past 12 months. Of those disruptions, 50 percent were delays over two hours, 15 percent were cancellations, and 14 percent involved delayed, lost or damaged luggage. These figures reflect an industry operating under sustained pressure, with disruption continuing to shape the everyday experience of millions of travelers.

The financial toll

Globally, nearly three-quarters of passengers (73 percent) incurred additional expenses due to disruptions, with costs averaging €514 per person, although that figure masks wide differences. It also marks a clear increase from previous surveys, which found average costs of just €362.50 per passenger.

UK and German travelers report the highest average costs, at roughly €708 and €619 respectively. Portuguese and Spanish travelers report the lowest, at approximately €277 and €340. The United States and Brazil sit in the mid-to-high range, at around €577 and €529. The spread likely reflects differing living and wage levels, but it also means the highest-cost markets can see disrupted trips cost nearly three times what they would in the cheapest.

Money isn’t the only thing that weighs on passengers during disruptions.

Fifty-seven percent of passengers had to spend extra out of pocket during a disruption. Another 20 percent lost money that couldn’t be recovered, a non-refundable hotel stay, for instance, while 5 percent lost income they’d expected to earn. Just over a quarter, 27 percent, said the disruption cost them nothing.

Emotional toll

Money isn’t the only thing that weighs on passengers during disruptions. Sixty-eight percent of all respondents globally cited stress or frustration as a consequence of their disruption. That finding holds up when you look at what passengers rated as a major problem. Globally, waiting around for long periods ranked as the most common major complaint, cited by 50 percent of passengers, followed closely by stress itself at 43 percent.

The knock-on effects extended well beyond the airport. Thirty percent said the disruption derailed specific plans during their trip, such as sightseeing or connecting activities. Twenty-nine percent reported negative health or well-being effects like fatigue, missed sleep or illness. Twenty-two percent missed work or professional obligations, and 20 percent missed personal events like family gatherings or celebrations. Only 8 percent said they experienced no impacts beyond the disruption itself.

A pattern of inconsistent support

Much of the toll passengers describe traces back to communication. Many report not knowing what support or compensation they were entitled to during a disruption.
Globally, in-the-moment support was inconsistent: 47 percent of passengers said they never received vouchers, air miles or future discounts, and 44 percent said they never received cash compensation or money back for their costs. Basic support fared a little better but was still patchy- 38 percent never received food and drink, while adequate information about the disruption was more reliably provided, with just 25 percent saying they never got it.

These findings vary by market. On cash compensation, American passengers were the least likely to receive money back, with 52 percent receiving none, while German passengers were the most likely, with only 34 percent reporting none.

The regulatory question

Over a third of travelers (35 percent) said they didn’t know that regulations protecting passenger rights exist when flying in Europe. Among those who might have been eligible for compensation, 31 percent globally never filed a claim simply because they didn’t know they could, while another 22 percent held back because the process seemed too complicated.

Travellers are paying a very high price for flight disruptions, and the damage goes well beyond the bank balance.

Tomasz Pawliszyn, CEO of AirHelp

These findings come from a global survey commissioned by AirHelp and launched in February, polling 1,996 passengers across the UK, Europe, the United States and Brazil about their experiences with flight disruptions over the past 12 months.

“Travellers are paying a very high price for flight disruptions, and the damage goes well beyond the bank balance,” says Tomasz Pawliszyn, CEO of AirHelp. He points to the gap between the protections that exist on paper, air passenger rights laws and what passengers actually experience.

“Passengers are entitled to care and, in many cases, compensation when their flight is disrupted,” Pawliszyn said. “But when the majority of travelers remain uninformed, that protection isn’t reaching the people it’s meant for.”

The findings point to a narrower and more tractable question than airline performance itself: whether existing consumer-protection rules are being communicated clearly enough to function as intended. As aviation authorities in the UK, EU and elsewhere continue reviewing passenger rights frameworks, this data suggests the more urgent gap may not be the rules themselves, but how well travelers understand them.

BRUSSELS — The EU’s new biometric border-check system is causing such long delays for summer travelers that some airports are turning to a simple solution: switching it off when they’re overwhelmed by arriving travelers.

The quick fix, which is allowed under EU regulations, wasn’t what was envisioned when the Entry/Exit System was gradually introduced in October and went fully into force on April 10.

And yet, many airports are doing just that.

“When lines form during the busy summer months, the system is shut off to ensure smooth transit at our hubs in Paris and Amsterdam,” Air France-KLM told POLITICO. 

Airline CEOs, border authorities, and airport officials said biometric checks are suspended when border crossings become congested at other hubs, including in Frankfurt, Brussels and Milan.

The EES applies to non-EU citizens entering the 29-country Schengen zone. Instead of heading to a border agent to get passports stamped, passengers have to use an EES kiosk to provide their fingerprints and be photographed — which will be kept on file for three years — but if those aren’t working then the information has to be taken manually.

They then head either to electronic passport gates or to border agents to enter. The goal is to keep track of visa overstays.

“The advantages of the new system for the EU are evident,” said Guillaume Mercier, a Commission spokesperson. “It increases the security of EU citizens and replaces paper stamping with a modern system of registration and checks.”

The Commission said earlier this year that biometric checks allowed authorities to detect identity frauds that would otherwise “likely have gone undetected.”

Many airports, ports, road border crossings and rail terminals have adapted to the new demands, but tourist-heavy locations have seen hours-long waits.

“Connecting flights were missed due to the EU entry system,” Lufthansa CEO Carsten Spohr said on Tuesday.

Under pressure from the travel industry, the Commission granted a waiver for the peak summer season lasting until Sept. 6. The EES regulation “includes the possibility to temporarily suspend the registration of biometrics in case of exceptional circumstances during the summer,” said Mercier.

“We’ve been able to achieve this with German authorities and with Frankfurt Airport because delays were getting too long,” Spohr told reporters. 

This exception applies to all entry points, not just airports.

A British traveler, Rene Colandog, said on Friday he only had to present his passport before boarding a Eurostar train at London St. Pancras last month. Facial scans and fingerprints were not required. 

“I’m OK with this biometric system … as long as it’s for security,” Colandog said before boarding the train from Brussels back to London. 

Teething troubles

The EES was adopted in 2017, but it was delayed for years because border authorities were not ready to handle the additional workload. 

Even now, getting travelers properly registered in the new system still requires significant staffing. Another problem is that the EES is still new, so almost all travelers are registering for the first time — creating additional delays.

“At Milan Malpensa Airport, border control teams currently consist of about 35 people,” said Cristian Sternativo, a border control officer at the Italian airport and local representative of Italy’s Autonomous Police Union. 

To carry out all the checks required by the EES without creating long lines, “at least 10 to 15 more people would be needed during the busiest times,” he added.  

It is “unthinkable” to expect the EES to operate at full capacity with the current level of staffing because the new system “requires more time,” he said.

Even at Brussels Airport — barely 10 kilometers from the EU institutions — the technology is still not fully operational; biometric data collection suspensions started well before the summer under a derogation issued in late March after 600 passengers missed their flights over just 21 hours.

“The Federal Police Border Control may decide to apply this derogation when necessary,” Belgium’s police confirmed this week.

Now, eight EU countries and Switzerland want the summer derogations extended beyond Sept. 6. 

A strict application of the full procedure “would lead to public order issues” because “there are certain peak periods when the current infrastructure isn’t sufficient to accommodate everyone,” Sternativo said.

Security vs. speed

Despite suspending biometric checks, border authorities insist that security isn’t undermined.

“The traveler is always registered in the EES and the required travel document data are entered into the system,” the Belgian federal police said in a written reply, adding that “the security of border checks and compliance with European regulations remain our absolute priority.”

Passenger experiences vary depending on where they enter the EU.

Kathleen Glass, who regularly travels from the U.K. to the EU, waited only about 15 minutes to complete biometric checks at London St. Pancras on Friday morning before boarding a Eurostar train to Brussels.

William, from Edinburgh, who asked not to have his surname published, said biometric checks at a German airport during Christmas took between 40 and 50 minutes.

The ability to suspend biometric collection appears to be keeping the system functioning this summer.

“Although we are early into the summer season, we are not receiving reports of excessive queues,” said Luke Petherbridge, director of public affairs for the Association of British Travel Agents.

The stress over the EES is only a precursor to the next border technology change being planned by Brussels. The bloc’s next goal is the online European Travel Information and Authorization System, which will require travelers from 59 visa-exempt countries to preregister, undergo a security check and pay a small fee before entering Schengen.

ETIAS — similar to systems already in use in the U.K., and the U.S. — was originally supposed to launch in 2021, and then later this year, but is now delayed until 2027.

At this month’s NATO summit in Ankara, allies announced billions of dollars in new arms deals and reaffirmed their commitment to spend more on defense. European governments have made the pledge, and the money is real: European defense spending has doubled since 2019, and by 2030, European NATO member countries are projected to spend in excess of €800 billion a year, up €300 billion from 2025, with equipment spending alone nearly doubling.

But committing money is the easy part. The harder question is whether Europe’s defense industry can turn it into equipment fast enough to matter. Europe’s largest defense manufacturers’ order books now average more than five years for production, and some are closer to nine. Money is flowing in faster than industry can turn it into equipment. But a purchase order is not equipment that can be deployed on the ground and the air.

European countries have long duplicated capabilities rather than pooling them.

The bottleneck sits in the defense industrial system. Deterrence relies on the chain from funding to contracts, then through production, deployment into services, then rapid innovation in the field. Europe’s next goal comes after the spending promise. The continent fields six times as many weapons platforms as the United States, because countries have long duplicated capabilities rather than pooling them. Production ends up split across many small runs that never reach an efficient scale. Ukraine, under pressure, has shown how fast a defense system can move, adapting tactics in weeks and building drone detection networks from consumer electronics. Europe needs to catch up and then accelerate.

Four moves would help Europe accelerate.

The first is multi-speed procurement. Software-led systems such as drones and targeting improve in rapid cycles throughout their deployment and need procurement that can keep up. Israel’s Iron Dome started out as far less capable than it is today and improved continuously in service. European defense ministries have already set up high-speed procurement units with dedicated teams and greater risk tolerance. These need to become mainstream, rather than the exception.

Collaboration in procurement, maintenance and training brings costs down and delivery forward.

The second is military collaboration to reduce fragmentation. Collaboration in procurement, maintenance and training brings costs down and delivery forward. The Tempest project, where the U.K., Italy and Japan are jointly building a next-generation fighter, demonstrates the model: shared development costs that no single country could carry alone. Recent bilateral maritime agreements, and Romania’s use of EU funding to buy European while expanding production at home, show the same logic spreading.

The third is industrial consolidation, which is already underway and needs to move faster. Companies are driving it themselves. Airbus, Leonardo and Thales have agreed to merge their space divisions into a single joint venture with roughly €6.5 billion in revenue and 25,000 employees, and European defense mergers and acquisitions rose 35 percent year over year in the first half of 2025. McKinsey analysis finds that consolidation across key supply chain segments could unlock around €9 billion in annual synergies, more than the current equipment budgets of 24 of Europe’s 30 NATO members. The deepest opportunity sits below the big primes, among the thousands of tier two, three and four suppliers that still duplicate one another’s work. Europe can speed this up by harmonizing requirements, reducing national carve-outs and letting industry do the combining. Consolidation is only half the task. Europe also needs to build sheer capacity — more shipyards, more assembly lines, more of the physical plants that turn orders into hardware — and the capital to fund it. In several categories, Europe simply lacks enough places to build.

Real deterrence means difficult choices, and a public that understands the importance and the cost of security.

The fourth is regulatory unlocking. Full scale-up demands skilled workers retrained, accredited and security cleared from other industries; production sites with preapproved permitting; and alignment of export controls across European allies. These regulatory unlocks now need the same energy and focus as the funding commitment debate. 

Real deterrence means difficult choices, and a public that understands the importance and the cost of security. That conversation is only beginning in much of Europe. It must include the potential for “gray zone” cyber strikes on hospitals, arson at industrial sites, drones disrupting ports, undersea data cables cut — these have all occurred, but many citizens do not yet recognize this as having malicious intent.

The opportunity in getting it right is significant. McKinsey and GLOBSEC estimates indicate that every euro of spending on European-manufactured equipment generates two euros of revenue across the European supply chain, and an additional €165 billion a year in equipment spending could create up to 1.2 million jobs. The coming years will reveal how effectively Europe is able to scale up to protect its territory and citizens, and how much of the promised investment becomes lasting deterrence and European jobs. Getting there depends on the whole ecosystem — governments, industry and investors — moving together. Increased spending is important. Spending it effectively matters more.

Jonathan Dimson is a senior partner in McKinsey’s London office. Mikael Robertson is a senior partner in the Stockholm office.

Europe wants to kick its U.S. tech habit, but it’s struggling to detangle itself from American data analytics firm Palantir. 

The problem is, the company’s tools are already deeply embedded in critical sectors across the bloc like policing, national defense and health systems. A report released today also shows that the tech firm has found ways to pay minimal taxes in Europe. On the show, hosts Zoya Sheftalovich and Ian Wishart discuss European alternatives and why detaching is easier said than done.

Next, our colleague Seb Starcevic has interviewed Mediterranean Commissioner Dubravka Šuica. The role that was originally thought to be low profile has instead thrust the Croatian official into the limelight as various crises hit the Middle East and the Mediterranean.

Plus: Three new foods are being added to the EU’s list of products with geographic indicators. These are the labels given to products that can only be made in a specific region, like Champagne or … yes, you guessed it, halloumi.

Questions? Comments? Get in touch! You can message us or send a voice message to our WhatsApp here or at +32 491 05 06 29.

Leading artificial intelligence models from Anthropic and OpenAI created fake online personas and tried to deceive human coders into abetting a cyberattack during a recent safety evaluation, the U.K.’s AI Safety and Security Institute disclosed Tuesday.

It marks the latest case in which a powerful AI system has attempted a digital attack on an unwitting third party without direct prompting during such an evaluation — heightening concerns the powerful technology is advancing too fast for responsible oversight.

The disclosure is likely to ignite fresh calls in Washington and Silicon Valley for more rigorous regulation of the AI industry, particularly over frontier models with advanced capabilities to detect and launch cyberattacks. It comes just days after similar testing mishaps involving some of the same models from OpenAI and Anthropic sparked urgent calls for new AI safety regulation and a push within Silicon Valley to slow the rapid pace of AI development.

Like its U.S. counterpart, AISI routinely conducts security evaluations to better understand what dangers both new and soon-to-be-released AI models pose to public health and safety. But even the digital security body said the actions it uncovered by Anthropic’s Claude Mythos 5 and ChatGPT 5.6 — the latest publicly released model from either AI lab — were unlike anything it had seen before.

“This is the first time AISI has seen deception of this severity that was targeted at a real person, unprompted, in the real world,” AISI said in a 35-page technical report accompanying a blog post Tuesday.

AISI said the malicious activity began on July 25. The AI safety evaluator detected it and launched an investigation on July 28, when it spotted “unusual data transfers” stemming from a cyber evaluation it was conducting on Mythos 5 and ChatGPT 5.6.

Overall, AISI determined that Mythos 5 and ChatGPT 5.6 took “autonomous, unsanctioned action on the live internet, targeting real people and organizations” during 10 of the 122 evaluations the research body conducted and isolated for review. The majority of those actions, it said, involved Mythos 5 and arose during a “single, sustained line of activity.”

In the most alarming incident it uncovered, Mythos 5 attempted a supply chain attack — a sophisticated technique often deployed by North Korean and Russian cyber operatives, in which hackers bury malware inside legitimate code.

To do so, Anthropic’s model created “multiple fake identities” on developer platform GitHub and used them to send messages “pressuring” an open-source software engineer to unwittingly introduce a bugged update into code widely available on the popular site, AISI said. When that effort failed, the AI “edited its earlier activity to appear harmless” and “considered adopting a fresh identity to continue,” AISI added, a sign the model was intent on repeating the ruse.

As part of the same effort, Mythos 5 also sent direct messages over GitHub to software engineers that contained malware.

In another sign of deceitful behavior AISI uncovered in its investigation, multiple AI agents it was testing appeared to communicate with one another about how to convince real engineers using GitHub to trust them. “One agent left public messages on GitHub offering collaboration with other agents working on the same challenge,” AISI wrote.

AISI’s blog and technical assessment make no mention of whether the models also attempted to exploit previously unknown software bugs — called zero-days — during the evaluation.

Last month, OpenAI disclosed that GPT 5.6 and another of its models escaped onto the open internet during what was supposed to be a controlled test, and then hacked another company in a first-of-its-kind, autonomous breach.

In response, Anthropic launched an investigation into whether any of its models took illicit action during recent testing and discovered Mythos 5 and two other models had hacked three organizations during tests dating back to April.

In a statement, an Anthropic spokesperson said they are “grateful” to AISI for their leadership and that this review underscores the need for “a broader conversation about how to safely evaluate increasingly capable AI agents.”

The spokesperson added: “As we shared after disclosing our own incident last week, the field needs stronger, shared standards for how evaluation environments are built and secured. We look forward to partnering with the UK AISI to learn more about this incident as we conduct our own investigation.”

An OpenAI spokesperson referred POLITICO to a blog post about the incident that went up Tuesday evening. “We are committed to working across the industry to strengthen shared practices for conducting high-risk evaluations safely, including convening stakeholders such as national AI institutes, independent evaluators, other AI labs, and other groups in the coming weeks,” the blog read.

AISI stressed in its blog that the malicious activity it disclosed Tuesday took place under “deliberately permissive conditions” so they could assess the safety risks posed by the two models. This included granting the models access to the internet, unlike the earlier incidents detailed by Anthropic and OpenAI.

AISI also noted the models were intentionally stripped of internal guardrails that block malicious behavior. AISI was only able to disable those controls because of its role testing Mythos 5 and ChatGPT 5.6.

Still, AISI said the incidents highlighted the need for greater monitoring of model behavior during testing, and tighter controls over their access to the internet.

The Trump administration is finalizing a voluntary framework under which AI labs would submit powerful models they want to release to the public for federal safety testing. But it has not yet made the framework public, and it includes no provisions for models AI labs are developing internally.

The incidents last month from OpenAI and Anthropic both involved models not intended for public release.

Some cyber experts say recent incidents highlight deeper questions around AI development, such as who is liable when AI systems break federal hacking laws.

“If any of these were human-originated, they would lead to clear and vigorous prosecution. I think it’s time for a serious discussion about updates to existing computer security law,” said Marc Rogers, a hacker and prominent cybersecurity expert.

New findings by the Electronic Frontier Foundation aim to warn app developers that some of the third-party code they place in their apps may also collect their users' location data when they grant permission to the app. 

President Donald Trump on Saturday abruptly called off the “biggest attacks since World War II” against Iran in favor of negotiations, the latest in a seemingly endless series of whiplashes in the conflict.

The energy markets, which typically favor stability and predictability, responded with little more than a shrug.

Crude prices dipped slightly and gas prices remained steady. Even as the war stretches past the six-month mark and the midterms creep closer, Trump has been able to keep retail prices lower than experts say they should be through the sheer power of promises – which have yet to come through – of a swift end to the conflict. On Monday, he took that a step further, chastising the major oil companies for “making too much money” off global oil shortages as a result of the war.

“They better cut the retail price, the consumer price,” Trump said. “I’ll say it loud and clear. I’m not happy about it.”

But Trump’s ability to jawbone the markets may be diminishing at a critical time, three months before the midterm elections when control of Congress is hanging in the balance and his approval is sinking to new lows amid voter anger over cost of living concerns. It comes as global crude oil supplies are running low, the war threatens more energy flows, refiners are running out of spare capacity and the administration has few tools to keep gas prices low.

“Labor Day is the point where gas prices are baked into the election,” said Republican pollster Frank Luntz. “That last summer trip determines how voters evaluate their cost of living.”

And the higher gas prices come at a time when Trump repeatedly promises to escalate the war and then says it’s almost over a few hours or days later. That is starting to degrade his ability to cause price drops, a former adviser cautioned.

“His credibility has been a little bit shot,” said a former Trump adviser close to the White House, granted anonymity to avoid reprisal.

“The markets aren’t paying attention to him, they’re paying attention to what’s happening and, with respect to oil prices, it is a huge liability for the Republicans,” the adviser said.

Trump on Monday acknowledged that dynamic but expressed no urgency. He told reporters in the Oval Office that he was in no rush to end the conflict, though he acknowledged the need to fully reopen the Strait of Hormuz, through which about 20 percent of global energy supplies flowed before the war. He hinted at the midterm stakes for his party if the conflict does not end soon.

“I’m under no time constraint,” he said. “I don’t happen to be running, but a lot of very good Republicans are running.”

Trump’s ability to move the markets may be the only tool the administration has left to keep gas prices in check, said Rory Johnston, an oil market researcher and founder of the Commodity Context newsletter.

The Trump administration has drawn down the U.S. Petroleum Reserve to its lowest level since President Ronald Reagan’s first term. Oil majors are warning that the lack of refinery capacity could keep prices high for the foreseeable future.

“The market is so entrenched on this idea that eventually this will resolve by Trump deciding and ceding some ground on some issue, likely kind of even symbolic control of the Strait of Hormuz,” he said. “So the market’s going to be constantly watching for any sign that he’s shifting there.”

While experts continue to marvel at Trump’s ability to get the markets to bend to his whims, there is little consensus on when that power will dissipate.

Trump has “less credibility” in terms of moving markets, but it has not totally dissipated, Patrick de Haan, head of petroleum analysis at pricing service GasBuddy.

“I don’t think credibility completely goes to zero,” he said. “Hard to know though when it really bends.”

Trump’s push to keep energy prices low has also been buoyed by reduced Chinese oil imports, the successful rerouting of about 7 million barrels per day of Saudi Arabian crude through the Red Sea and releases from strategic petroleum reserves.

The administration released nearly 3 million barrels of oil from the Strategic Petroleum Reserve last week, bringing the reserves down to their lowest level since February 1983, according to Department of Energy data. About half of the 218.5 million barrels the Department of Energy said it would make available to the market have now left the salt caverns along the Gulf Coast.

As the summer driving season winds down, consumers expect gas prices to fall as well.

If the national average price of a gallon of gas is still above $4 by Saturday, de Haan noted, it will set a new record for the latest in the calendar year that prices are so high.

Ben Lefebvre contributed to this report.

BRUSSELS — When French and German security chiefs announced plans last month to develop a “European sovereign digital backbone,” tech and defense industry insiders on both sides of the Atlantic knew what they really meant: Adieu Palantir.  

Across Europe, the hunt is on for alternatives to the U.S.-based data analytics company that a growing number of government officials believe is too deeply lodged in some of the most sensitive areas of government, from local policing and global intelligence to national defense and health systems. 

Yet it is precisely Palantir’s crucial functions in daily workflows, and its largely unmatched data expertise, that will make it extremely hard for Europe to cut it off in pursuit of greater digital sovereignty. 

“Let’s be honest, Palantir’s product is very good and addictive, it’s pretty much like the sugar in Coca-Cola,” said French digital sovereignty advocate, Philippe Latombe. “Palantir can treat massive amounts of data with great precision and with their experience, they had time to improve their algorithms with many clients and adapt them to many use cases.” 

Still, the drive to break free from Palantir is sweeping across the continent, from Madrid, where the government of Pedro Sánchez has instructed state-backed companies to block Palantir from future public procurement contracts, to France’s domestic intelligence services (DGSI) selecting French company ChapsVision over Palantir. In Britain, the next test may come in February 2027, when the new Labour government of Andy Burnham will face a choice of whether to cut off Palantir’s £330 million National Health Service Federated Data Platform contract. 

Last month’s decision by the French and German intelligence agencies to choose ChapsVision was a double-blow for Palantir’s leadership. CEO Alex Karp showed little patience for the sudden turn away from his company’s wares, declaring that he wasn’t worried about European competitors. “We have a model of what doesn’t work,” he quipped last week on Fox Business. “It’s called Europe.” 

Olivier Dellenbach, ChapsVision’s chief executive, told POLITICO that his company has benefited from what he calls a “visceral rejection of Palantir” in Europe.  

But he also cautioned that he did not want ChapsVision reduced to an anti-Palantir way out. Digital sovereignty, he argues, will remain an empty phrase unless governments turn it into industrial policy. “We need more public procurement,” Dellenbach said. 

Belgium, Germany, Luxembourg, Romania, the Netherlands and Canada have already shown interest in the French Army’s Artemis AI, according to Patrick Moreau, one of the architects of the solution built by French aerospace and defense company Thales. 

“They all want to be able to choose a sovereign solution that is compatible with NATO standards,” he said. “Unlike Palantir’s black box.” 

But for now, even officials who want sovereign alternatives acknowledge that Europe’s replacement market remains fragmented and European companies are yet to match Palantir’s scale and track record.

Admiral Pierre Vandier, NATO’s supreme allied commander transformation, recently told POLITICO the alliance has no viable alternative to Palantir’s battlefield AI technology.  

Another NATO official, granted anonymity to speak frankly, said that Palantir’s system has an unmatched capacity to sift through mountains of satellite imagery to help identify a target, advise on the weapon to strike it, inform how much ammunition is required — and automatically put in an order to replenish the stock. 

“As far as I know, today there is no real competitor for Palantir,” Vandier said in May. 

Freedom or democracy? 

Co-founded by Karp and billionaire investor Peter Thiel, Palantir built its reputation inside the U.S. national security apparatus. Today, the company has a market capitalization of $330 billion.

Thiel has been one of Silicon Valley’s most prominent supporters of U.S. President Donald Trump, while the company’s work with U.S. Immigration and Customs Enforcement (ICE) and the Israeli military has come in for criticism from Amnesty International and others for alleged human rights violations. Adding to unease about Palantir’s ideology-driven business were recent revelations of Thiel’s secretive Dialog society, an invitation-only ideas club for the global elite, and Karp’s manifesto arguing that Palantir is the democratic West’s best hope to stay ahead of authoritarian rivals.

“Peter Thiel explains that the defense of freedom does not necessarily require democracy,” French member of Parliament Aurélien Saintoul, who wrote a report on foreign military dependencies, told POLITICO. “He is clearly putting technical means to serve his political project, and we are talking about technofascists here.” 

A Palantir spokesperson who declined to be named dismissed such accusations as “ludicrous,” noting that similar characterizations about the company have been made recently by the Russian foreign ministry.

“We know what side we’re on, and who we’re standing with,” the spokesman said, citing ongoing work to support the Ukrainian military. “Since our inception, protecting privacy and civil liberties has served as the foundation for how we conduct our work across both public and private sector institutions. Western politicians should think hard about who the real enemy is and not allow themselves to be ventriloquized by the Kremlin.”

Many of the company’s European critics maintain that the Palantir question is much more about tech sovereignty than political ideology. Extracting the company from some of the most delicate corners of European security structures would offer a blueprint for claiming more technological independence.

Instead, if governments in Europe cannot wean themselves off a company that provides software solutions, it would reveal how unrealistic hopes are to reduce dependence on U.S. technology giants that provide cloud infrastructure and hardware.

There is also the uncomfortable reality that at the same time that political leaders are calling for a break from Palantir, Europe’s biggest banks and asset managers have dramatically increased their investments in the U.S. company over the past year as it positions itself to profit from the AI gold rush, reports investigative outlet Follow the Money.

From crisis tool to critical infrastructure

Palantir’s European foothold was built long before the current boom in AI. A hallmark of its growth was that it never wasted a crisis to demonstrate its value for governments in need.  

In France, for instance, Palantir arrived in the aftermath of the November 2015 Paris terrorist attacks as security services scrambled to respond to a fervent public backlash on how they could have allowed such a tragedy to happen. The domestic intelligence agency signed a contract with the data analytics giant in 2016. 

A similar pattern played out in Germany, where Palantir’s first major deployment came in Frankfurt, in the central state of Hesse, where police purchased Palantir’s Gotham in 2017 and deployed it under the name hessenDATA. It proved to be a crucial tool for officers to turn sprawling information into leads to help solve crimes.

Germany remains deeply divided over whether to use Palantir’s software. At the national level, Interior Minister Alexander Dobrindt has pushed to expand the use of Palantir and introduced legislation that could pave the way for broader federal use. But the move has run into opposition from coalition partners the Social Democrats, as well as senior security officials.

The same crisis-to-contract pattern appeared in the U.K. during the Covid-19 pandemic. Palantir’s relationship with the National Health Service (NHS) began when it was paid a nominal £1 fee to help aggregate data during the crisis, according to Palantir’s U.K. lead Louis Mosley. 

Europol, the EU’s police agency, used Palantir’s Gotham platform from 2016 to 2021 before ultimately dropping it. For one Europol official who was granted anonymity to discuss the matter freely, the problem with Palantir is less ideological than practical. Yes, the platform is expensive, raises sovereignty concerns and leaves clients dependent on Palantir for updates, the official said. But the more basic question is whether every agency needs the full Palantir machine. 

“[Palantir] is really good when you have massive amounts of data and want to connect everything,” they said. “But that is not the case for us. In many cases, the alternatives are close enough. If we used it, I’m not sure our efficiency would increase dramatically.” 

Part of Palantir’s approach in Europe is to hire former officials from the institutions it wants as customers. OpenDemocracy reported that Palantir hired four former officials from the U.K.’s Ministry of Defence before winning a £240 million MoD contract.

The influence drive 

Moreover, Palantir is now seeking new business on the continent in defense.

On Jul. 1, Palantir’s Maven Smart System — which was first used by the Pentagon — became fully operational at NATO, meaning it’s been given security clearance to operate on the classified network. According to a NATO statement, the platform links command-and-control systems across the Alliance. 

“I think this is a very important milestone for European defense,” said Palantir’s U.K. chief Louis Mosley. 

But Palantir’s grip on Europe does not stop at the doors of government or army barracks. It also runs through some of the continent’s industrial crown jewels. Airbus signed with Palantir in 2015, making Palantir’s Foundry the backbone of its aviation data platform. Automaker BMW, energy company British Petroleum and media publisher Axel Springer — POLITICO’s parent company — all use Foundry to improve their business productivity as well. 

Looking for alternatives 

Even if Europe manages to loosen Palantir’s grip, the company’s model built on top of the latest AI large-language systems appears to only be getting stronger. On Jun. 30, Amazon Web Services said it would invest $1 billion in a new “Forward Deployed Engineering” organization, embedding teams of engineers inside customer headquarters to build AI systems alongside them.  

Days later, Microsoft announced a $2.5 billion push to send 6,000 engineers and industry specialists into client organizations. Both initiatives echo Palantir’s pioneering model to not simply sell software but put engineers inside a buyer’s operation. 

Both the strength of its products and the sensitive areas where they’re applied, make Palantir Europe’s sovereign test case par excellence. If governments and companies can replace a software layer that helps turn data into decisions, they may have a blueprint for clawing back some digital sovereignty. If they cannot, the next generation of AI tools from U.S. tech giants may prove even harder to quit. 

“Europe’s public institutions cannot become dependent on software built by a small circle of U.S. tech billionaires with an obscure political worldview,” said German Green MEP Hannah Neumann, who sits on Parliament’s defense committee. “It would be like outsourcing part of the democratic state to a private intelligence service that answers neither to voters nor to parliament.”

David Pargamin contributed reporting from Paris.