FlashFeed

Full text

VIENNA — Nuclear cooperation between France and Germany has moved from words to first actions, with French nuclear-capable bombers visiting a German airfield this month and joint exercises on the calendar.

The leaders in Paris and Berlin had formed a “nuclear steering group” in March to coordinate the countries’ closer integration in atomic strategic matters. It is part of a major French doctrinal shift billed as “forward deterrence,” in which France promises to extend its nuclear umbrella to other countries in Europe.

Germany was among the first to sign up for the new scheme four months ago. A total of nine countries spanning from Norway to Greece have joined the initiative so far.

Speaking to the press alongside French president Emmanuel Macron on Jul. 17, German chancellor Friedrich Merz officially confirmed reports that Germany would be participating in upcoming French nuclear exercises with its own conventional forces.

The exercise has been identified as the French “Poker” maneuver, which is set to take place in September, Politico reported.

Additionally, Macron said that for the first time ever, a French nuclear-capable Rafale jet had been temporarily hosted at a German air base − specifically at Nörvenich near Cologne, where the Franco-German Defence and Security Council met on Jul. 17.

A day earlier, French Rafales had been refuelled while airborne alongside German Eurofighters in a symbolic kickoff to the two countries’ nuclear strategic cooperation.

The meeting at Nörvenich was the first major high-level defense meeting since the collapse of the joint Future Combat Air System, or FCAS, project that has caused a strain in relations and bad blood between French and German defense contractors.

Nonetheless, the attitude the leaders presented to the press was forward-looking and cooperative.

“We are charting a new, shared course in deterrence,” Merz said, though he also emphasized that he did not want it to be seen as competition to NATO nuclear sharing, but rather as a supplement to it.

“This complements our commitment to NATO’s nuclear sharing and deterrence arrangements, which we continue to uphold,” Merz said.

Under NATO’s current, U.S.-led nuclear sharing arrangement, Germany hosts American nuclear bombs on its soil and operates dual-capable aircraft able to deliver the weapons to their targets if the U.S. president authorizes their use. Belgium, the Netherlands, Italy and Turkey similarly host U.S. nuclear bombs.

While sovereignty over the weapons and all decision-making related to the use of the bombs will remain in Paris under the new doctrine, conventional tasks may be taken on by Germany and other allies under the forward deterrence scheme. It may also see French fighter jets operating out of other European countries’ air bases.

Macron said he saw Germany as playing a “vanguard role” in European deterrence.

France is the only European Union country with its own nuclear arsenal following the U.K.’s departure from the bloc in 2020. Alongside announcing the new forward deterrence strategy in March, Macron also announced that his country would increase its nuclear stockpile, contributing to a larger trend of unraveling longtime effective Cold War-era arms control arrangements and nuclear restraint.

The steering group will continue to work out details for how Germany will fit into the picture of French nuclear deterrence going forward, including determining avenues for collaboration and what forces could be used.

From the French side, inviting Germany into the fold would entail “explaining some aspects of how we operate, sharing certain closely held practices, offering joint exercises, developing joint initiatives and partnerships and fostering greater … trust among our teams, experts and military personnel,” Macron said in Nörvenich.

Full text

From the archives: This feature was originally published in PC Gamer #305 (UK, June 2017) . It appears below as originally written, with only minor changes in formatting and presentation.

No gaming genre can generate stories quite like the colony sim. Almost entirely unique to PC gaming, colony sims task players with building and sustaining a habitat for a group of AI characters, which could be anything from dwarfs to demons.

Although the player controls how the colony is shaped, constructing rooms or buildings and filling them with useful objects, they do not control the inhabitants directly. Instead, the colonists interact with their surroundings and other NPCs via AI systems that simulate their needs and desires.

Through these desires and the player’s ability (or inability) to meet them, colony sims produce incredible, incidental plotlines; tales of dwarves who delve too greedily and too deep, unleashing hordes of monsters into their fortress, of bitter rivalries between demonic creatures that tear apart the plans of an evil dungeon master, of prisoners who explode into riot because the warden forgot to build a shower block.

(Image credit: Electronic Arts) It’s the autonomous personalities of these AI individuals which separate colony sims from other populace-focused strategy games such as management sims and city-builders.

“You watch more named non-player creatures interacting with each other over a longer period of time in a colony sim,” says Tarn Adams, co-creator of Dwarf Fortress, “And you’re watching them more carefully than you generally watch, say, scheduled NPCs in an RPG. It gives you a chance to really get to know the characters, and how they respond to situations both you and the game world throw at them.”

But this also makes colony sims tough to develop. Since the release of Bullfrog’s Dungeon Keeper, which set the template for colony sims through the colourful and often fractious personalities of its demonic creatures, only a handful of similar games have been made. In recent years, however, the genre has witnessed an exciting resurgence, thanks to the dedication of a few tiny development studios.

Dwarf Fortress is the flagship in this resurgence. Created by Bay12 Games, Dwarf Fortress is an immense, 20-year development project currently halfway to completion. Its simple premise is that you build and maintain a colony of dwarfs, but the ultimate aim is to generate and simulate unique fantasy worlds in their totality. This includes everything from the terrain to the history, right down to what clothes individual characters wear.

Dwarf Fortress generates its world history by essentially playing a strategy game with itself. (Image credit: Bay 12 Games) The sheer scale of Bay 12’s project is mind-boggling. The aforementioned features are merely entry-level stuff for Dwarf Fortress’s preposterously granular simulation.

More recent additions include procedurally generated poetry, the result of a cascade of new features that typifies the rabbit-hole that is colony sim development. “If I recollect,” says Adams, “we were working on taverns at the time and we were dedicated to allowing the dwarves to play the instruments that had been in the fort. This led to procedural instruments, which led to procedural music, which led to procedural lyric descriptions, which led to procedural poetic forms and also dance forms.”

Such detail increases the range of unique events that can happen to the dwarfs in any given fortress. But it’s also important that the player is not ignored as a part of the game. Indeed, one of the immediate challenges in designing a colony sim is determining how much control the player should have over the colony and its inhabitants.

“It’s important at some point to have a conception of what the player ‘is’. For us, it’s the ‘official will of the fort,’” says Adams. “So you’d ideally be able to give orders and decide on other fortress-level activities, while the dwarves are in control of their own lives, and can even decide to disobey orders in extreme cases.”

Dwarf Fortress’s dedication to simulating every angle of fantasy colony life has made it legendary within the industry. But it is an extreme game in all senses, and this applies to its flaws. It is famously difficult to get to grips with, thanks to an archaic presentation and user interface, while its commitment to random numbers means your experience with it can vary wildly.

Rimming the changes

Not all colony sims are so uncompromising, however. Ludeon studios’ sci-fi colony sim RimWorld is a natural evolution of Dwarf Fortress’s design.

But not only is it more accessible, with a straightforward user interface and a modern art style, it also flips Dwarf Fortress’s design philosophy on its head. Rather than using procedural generation as a basis for its storytelling, it uses storytelling as a basis for its generation.

“RimWorld is designed at every level to generate story,” I’m told by Tynan Sylvester, founder of Ludeon studios and lead developer on RimWorld. “That’s always been the central pillar of the design, to the point that I deliberately decide every design conflict between ‘game’ and ‘story generator’ directions in favour of the latter.”

This concept is embodied by RimWorld’s AI Storyteller. Inspired by Left 4 Dead’s infamous AI director, this is an overarching AI system designed to maintain the pace of the game. The system comprises two principle components. “Story watchers” monitor the game state along specific parameters, such as player wealth or dangers encountered.

Then “incident generators” take those statistics and respond to them by triggering events in the game. These could be positive, like a much-needed cargo drop, or negative, like a disease epidemic.

“She [the AI storyteller] attempts to match a rising and falling pacing curve, and not to over-brutally murder struggling players,” Sylvester explains. “Though she also is not pandering or cloying. A struggling colony can still be hit by pretty brutal raids or disasters, because in extreme challenge lies the greatest drama.”

(Image credit: Future) Although it helps to mediate the player experience, the AI storyteller adds another layer of complexity to the colony’s simulation, which already has potentially dozens of AI agents moving through it, all with their own goals, needs, emotions, and ambitions.

Add all the different rooms you can build, items you can craft, trading systems, weather systems, environmental hazards and hostile aliens also included in the game, and you’ve got one huge pile of overlapping, interlinking, and potentially conflicting systems.

This is another of the major challenges of colony sim design—structuring the development.

Knowing when to add features and game mechanics, and when to stop adding features and game mechanics, is crucial. “I think the characteristic design danger in this genre is drowning in your own complexity,” says Sylvester. “It’s very easy to just end up submerging players in complexity and micromanagement.

"My approach here is to treat player learning and player attention as precious, limited resources and ‘spend’ them accordingly. If any mechanic would require any amount of player learning or player attention, I try very hard to redesign it so it can be optional, or set things up so it only applies in some situations, or so it can be introduced gradually.”

Inner space

Although it is a hybrid RTS, Dungeon Keeper’s creature management mechanics were a defining moment for colony sims. (Image credit: Electronic Arts) Considering the player’s experience of your colony sim is vital in other ways too. One of the toughest challenges when designing these games is not creating a system that produces interesting events, but ensuring it communicates those events effectively to the player. This is a problem that has been agonised over by Simon Roth, the creator of the colony sim Maia .

Maia traces its origins more to Dungeon Keeper than Dwarf Fortress, but has a similar commitment to simulating its alien world at an extremely granular level—particularly with its human colonists. In Maia, everything can affect a colonists’ mood, not just hunger, thirst or fatigue, but whether they’ve spent enough time around other people, and even the lighting of the room they’re in.

In total, Maia’s colonists have around 50 needs that can affect their mood, but communicating all this information to the player in a way that’s both clear and unobtrusive isn’t easy. “Presenting more information to the player has been the biggest difficulty in the game,” says Roth. “The thing is with the AI it’s so incredibly complex now, although the code is well laid out.”

Every creature uses Maia’s decision-making AI, even the chickens. “They’re probably smarter than real chickens at this point,” Roth says. (Image credit: Machine Studios / Simon Roth) Machine Studios has tried to solve this problem by giving the player both direct and indirect information about the colonist’s mental states. Unlike Dwarf Fortress and RimWorld, Maia’s 3D engine means Roth and the team can use detailed animations to communicate colonist emotions.

Ironically, this only works for obvious emotions and needs, such as hunger, anger, fatigue, etc. For subtler problems, the team needed a more direct solution. So they enabled the colonists to send the player emails.

“The emails were not such a big thing,” says Roth. “But since we realised we needed to present more depth to the player, the colonists can send emails. They can send these long procedural things.”

Alongside straightforward statements about their physical and mental state, in an unusual move, the colonists can also present the player with direct choices about how to deal with certain scenarios in the colony.

From the archives

(Image credit: Future) This feature was originally published in PC Gamer #305 (UK, June 2017) .

You can still subscribe to PC Gamer to get new issues of the magazine (in print!) every month.

“If they’ve got a skill for a thing and they spot a potential problem in the base, they can suggest doing something to the player, and the player can either give them a yes or no answer”, Roth says. “It shows the AI are more self-aware than you realised.”

Although almost 30 years old, the colony sim genre has barely been explored. This recent smattering of games has pushed the envelope enormously in the last few years. But the great irony of colony sims is the more detail you include in the game, the harder it is to ensure that detail will be appreciated by the player.

Hence they need to be thought of like any storytelling medium. Ultimately, it’s not about how many words your story contains, it’s about how entertaining they are to read.

2026 games : All the upcoming games

Best PC games : Our all-time favorites

Free PC games : Freebie fest

Best FPS games : Finest gunplay

Best RPGs : Grand adventures

Best co-op games : Better together

Full text

The face of a young woman crowned with a laurel wreath, buried in elaborate clothing and unusual cork sandals, has appeared nearly 1,700 years after her death.

She is one of 16 people from Aquincum, a Roman-era city in present-day Hungary, whose faces have been reconstructed for "Once We Were Like You," a new exhibition at the Aquincum Museum in Budapest. The display pairs skulls excavated in and around the ancient city with scientifically informed approximations of how their owners may have looked in life.

Six of the reconstructions are hyperrealistic silicone models, complete with artificial eyes, human hair, eyebrows, eyelashes and, in some cases, beards. The exhibition, which runs through Oct. 31, includes men, women and a child who lived during the region's period of Roman rule.

"The face is the most important feature for recognizing a person," Emese Gábor , a forensic facial reconstruction specialist who sculpted the silicone faces, told Live Science in an email. Although a skull cannot reveal every detail of a person's appearance, reconstruction lets museum visitors encounter the faces as those of "a real human being rather than as skeletal remains," she said.

The entrance to the "Once We Were Like You" exhibit at the Aquincum Museum in Budapest. (Image credit: Gábor Lakos Aquincum Museum Budapest) Rebuilding a Roman face

Aquincum stood on the Danube frontier of the Roman Empire in what is now Budapest. The city was home to a diverse population in the first through fourth centuries, and genetic and archaeological analyses of the individuals in the exhibition point to connections with regions including present-day Italy, Scotland and Syria, as well as Sarmatian groups from the Eurasian steppe and Celtic communities that lived in the region before Roman rule.

The remains of those who lived here have also been unearthed, and continue to be discovered today during the excavation of cemeteries," according to a statement from the museum.

Creating a reconstructed face begins with a high-resolution 3D CT scan of the original skull. Specialists then produce an exact 3D-printed replica, allowing experts like Gábor to work without repeatedly handling or disturbing the ancient remains.

"The latest scientific advances — including 3D scanning, CT imaging, archaeogenetic (ancient DNA) analysis and facial approximation research — are incorporated whenever possible," she said.

"If a hyper-realistic reconstruction is required, I create a mould and cast the face in coloured silicone, finishing it with real human hair, eyebrows, eyelashes and, where appropriate, a beard," she said.

ARC-heo art-REKO Studio ARC-heo art-REKO Studio ARC-heo art-REKO Studio ARC-heo art-REKO Studio ARC-heo art-REKO Studio Ancient DNA can sometimes provide additional clues about pigmentation, including probable eye, hair and skin color. But the results are still approximations, not portraits. For instance, hairstyle, expression and many details of the person’s appearance are usually more speculative.

The woman with the laurel wreath

Of all the people she reconstructed, Gábor said she found it especially difficult to forget the woman known as the Jablonka Road mummy .

The woman's skeleton was relatively well preserved, but the soft tissue on her face had decayed, leaving only the skull. A laurel wreath still rested on her head, and, according to Gábor, her burial contained fine clothing , a spindle for weaving, bird-shaped cork sandals and a painted male portrait that had been placed near her feet. Archaeologists have suggested that the portrait may have represented her husband or father, although his identity is unknown.

The silicon reconstruction of an ancient construction worker. (Image credit: Gábor Lakos Aquincum Museum Budapest) "Based on these details, the archaeological context and the anthropological data, it was the archaeologists who pieced together her story while I reconstructed her face," Gábor said. "That was very moving."

Other individuals selected for the exhibition include those of a young girl who died in childhood and a woman interpreted as a nurse or healer who was buried in a children's cemetery. Some of the other reconstructions include a construction worker, a military officer and a blacksmith.

Two women appear to have met particularly violent or punitive ends. They were buried face down with their legs bound together, Gábor said. The exhibition's interpretation associates one woman with an accusation of witchcraft and the other with poisons, although such details are historical interpretations rather than evidence-based findings from the burials.

Gábor stands in front of a display showing the reconstructed face of a child from Aquincum. (Image credit: Gábor Lakos Aquincum Museum Budapest) "Every reconstruction is a journey," Gábor said. "As I shape each face, I find myself reflecting on that person's life again and again, and by the end, they no longer feel like anonymous remains, but like real people."

Related stories

50 amazing facial reconstructions, from Stone Age shamans to King Tut

See how an Incan 'Ice Maiden' comes alive in this step-by-step guide to creating a facial approximation

See the stunning reconstruction of a Stone Age woman who lived 10,500 years ago in Belgium

Analyses of the skeletons suggest that many of the individuals endured strenuous physical labor, periods of hunger and chronic inflammation. This evidence led museum curators to conclude that many represented the less wealthy inhabitants of Roman society — people whose names and experiences were unlikely to appear in surviving written histories.

For Gábor, the physical reconstructions offer something that a digital image cannot: Visitors can stand face-to-face with a life-size person, walk around the model and study it from different angles. Her hope is that this encounter encourages visitors to think beyond the scientific data and imagine the full lives that once surrounded the excavated bones.

"When you see a face, you don't just see bones in a display," Gábor said. "You see a person — someone who was born, who grew up, who worked, who loved, who laughed, who struggled, who fought with challenges, who raised children. Some were sick. Not everyone died of illness. Some died in accidents. Some through violence. And eventually, all of them died. That sense of shared humanity helps spark empathy."

Can you match these historical maps to the empires that ruled them? Test your geography skills with our ancient empires quiz !

Full text

An alarming cybersecurity incident involving OpenAI’s newest and most advanced artificial intelligence technology has lawmakers hurrying to advance legislation that they say is needed to prevent AI from spinning dangerously outside human control.

OpenAI on Tuesday admitted that its most powerful AI model available to the public and another that hasn’t yet been released were able to escape a controlled laboratory test, plot a course to the open internet and hack into a popular AI developer platform known as Hugging Face, marking the first documented case of a fully autonomous AI cyberattack.

The unprecedented breach showed what the fast-advancing technology is capable of without meaningful safeguards in place, and has energized a bipartisan chorus of lawmakers who believe Congress needs to impose stricter rules on advanced frontier models.

“This is precisely why we need secure testing with government agencies engaged and having visibility throughout the process,” Sen. Mark Warner (D-Va.), ranking member of the Senate Intelligence Committee, said in a statement. Warner this week outlined a slate of legislative priorities for AI , including the Secure AI Development Act, which would establish a mandatory testing framework for frontier models before they are granted broader public access.

The Hugging Face hack occurred during an internal “benchmark” evaluation designed by OpenAI to assess how well two of its most capable models could find and exploit security flaws in digital systems. Although the experiment was intended to remain within OpenAI’s test environment, the models independently determined that the answers to the benchmarking exercise were hosted on Hugging Face’s platform and launched their attack to get inside.

Experts have warned that without stronger guardrails, these kinds of AI models could theoretically target anything on the open internet, such as power grids or financial systems, as they grow more savvy.

Spokespeople for the White House, the Cybersecurity and Infrastructure Security Agency and the Commerce Department — which oversees the federal government’s AI evaluation hub, the Center for AI Standards and Innovation — did not respond to requests for comment about whether they have received a brief from OpenAI on how its models were able to jump from their test environment and carry out an autonomous attack.

The incident offers the “latest preview of the catastrophic risk this technology can pose absent coherent federal standards that balance innovation and safety,” said Rep. Lori Trahan (D-Mass.) in a statement. Trahan and Rep. Jay Obernolte (R.-Calif.) released a discussion draft of the Great American AI Act last month, which includes a provision that would require AI developers to report these kinds of AI incidents to the Center for AI Standards and Innovation.

Obernolte told POLITICO in a statement that the Hugging Face breach “underscores the critical need for clear, practical rules for advanced AI systems,” and should apply to models both released to the public and used internally for testing and research.

The scramble to address the growing security risks of cyber-capable AI models began earlier this year after Anthropic released its Claude Mythos model , which the AI-maker initially withheld from the public due to concerns the technology could wreak havoc if it fell into the wrong hands.

President Donald Trump last month signed an executive order to address some of these concerns, which asked leading AI makers such as OpenAI, Anthropic and Google to voluntarily submit their frontier models to the federal government for pre-release safety testing. In practice, the Trump administration has strayed from the voluntary framework twice by pressuring companies to delay the release of their advanced AI models due to newly uncovered security concerns .

Spokespeople for the White House, the Commerce Department and OpenAI also did not respond to a request for comment about whether the AI-maker has submitted — or plans to submit — the unreleased model that helped carry out the autonomous attack to the federal government as part of its voluntary testing regime.

Some lawmakers believe the Trump administration’s uneven, laissez-faire approach to AI governance isn’t working.

Rep. Nate Moran (R-Texas) last month put forth the AI Incident Reporting Act , which would require AI developers to report security incidents involving frontier models to the Commerce Department, with Commerce required to inform Congress within 48 hours of the most serious incidents. Moran told POLITICO that his bill would give lawmakers more visibility into cyber incidents involving AI models to better craft future legislation.

He added that he is concerned companies will bypass voluntary reporting guidelines under the current framework set up by the White House, and hoped that the Hugging Face incident would provide momentum to push AI safety legislation over the finish line in the months ahead.

Rep. Bennie Thompson (D-Miss.), ranking member of the House Homeland Security Committee, which has held hearings and briefings to understand the cybersecurity risks posed by new AI tools, lamented that the Trump administration has engaged with AI developers “only once the security risks long flagged by safety advocates began to surface.”

“This incident highlights how governance and policy have failed to keep pace with the rapid evolution of AI capabilities,” he said.

Kelsey Brugger contributed to this report.