Mythos helped to find a new meet-in-the-middle technique that relies on a Möbius Bridge, a more sophisticated fingerprinting algorithm used in meet-in-the-middle attacks. Using it, Green said, the code Mythos produced was able to reduce the number of required inputs to 2⁸⁹. Anthropic said that savings can reduce the time required for such attacks by 200- to 800-fold.

The ability to produce that many inputs makes the attack beyond reach outside of the laboratory. Further, the actual speed-up is unknown, since the weakened AES algorithm tested used only 7 rounds. Specification-compliant AES, Green said, uses 10, 12, or 14 rounds, depending on key size.

Anthropic is careful to explicitly spell out most of these caveats. The Monday blog post goes on to argue, however, that the results are nonetheless meaningful and could ultimately fundamentally disrupt the process of cryptanalysis, or the adversarial testing of cryptosystems.

“The cybersecurity community is now grappling with the fact that language models are able to discover so many bugs that the standard human processes (like vulnerability triage, verification, and remediation) struggle to keep up,” Anthropic wrote. “We predict that the same will soon be true in academic cryptography research. As language models increasingly produce novel research outputs autonomously, human researchers may become bottlenecked on studying and validating these results for technical validity, novelty, and utility.”

Not mentioned in Anthropic’s report is whether its researchers used Mythos to attack more tested cryptosystems, such as elliptic curve cryptography and RSA. Attack improvements against these systems would be more impressive. By achieving the most impressive result against an algorithm still in its infancy, it’s not clear how much of an advantage Mythos truly provided. There’s no way of knowing if researchers using conventional cryptanalysis techniques were already close to discovering the same attack.

Ultimately, the lesson from the research is simple. AI-assisted cryptanalysis remains untested, and providers of these platforms have a vested interest in exaggerating their benefits. At the same time, there’s growing evidence that LLMs may provide significant advantages in finding cryptographic weaknesses. It would be a mistake to conclude that LLMs won’t one day play an important role in the race between securing and compromising our most vital assets.

The headline and body of this story have been updated to reflect the withdrawing of HAWK.

The round builds on the company’s $4.2m pre-seed round in 2025 and will be used to scale and grow the organisation’s quantum ambitions.

Zuriq, a quantum computing company based out of Switzerland, has announced a seed funding round of $25.5m. 

The round was led by Quantonation, with participation from Forward.one, Extantia, Firgun Ventures and additional previous round investors. The company raised a $4.2m pre-seed round in 2025.

Zuriq is a spin-out of ETH Zürich and intends to use the funds to scale its trapped-ion quantum processors and expand its team alongside research and development efforts as it works to significantly increase the number of qubits its architecture can support. 

According to the organisation, the majority of trapped-ion quantum computers are based on a decades-old design in which ions are held in one-dimensional chains stitched together into larger grids using complex junction structures. 

Zuriq claims to have taken a novel approach with processors that are natively two-dimensional and infrastructure that allows ions to move freely in any direction without the junctions that constrain reconfiguration.

Prof Jonathan Home, a scientific adviser at Zuriq and professor of the Department of Physics at ETH Zürich, said: “What makes this approach powerful is geometry. Hold ions in a line and the count grows one at a time, hold them in two dimensions and it grows with the area of the chip, on a standard chip, that is the difference between tens of ions and many thousands. 

“Just as important, the ions can be moved freely in three dimensions, so they can be connected together far more flexibly, and that connectivity is what ultimately makes a quantum computer more capable. Reaching the scale and connectivity that real applications will need has been the central challenge for the whole field.”

Zuriq has stated it is now turning its attention to scaling its technology and plans to increase the number of ions on a chip to enable powerful, commercially useful machines. 

Christophe Jurczak, a founding partner at Quantonation, said: “A common misconception is that the quantum race is already decided, but that is far from reality. Zuriq is demonstrating that there remain significant and transformational physics breakthroughs still to be made in quantum architectures. 

“The scientific calibre of the team is world-class, which is why Zuriq has been phenomenally successful in attracting global talent to join them and build 2D-native quantum computers where trapped-ion qubits finally have the freedom to move and connect.”

Also this week (27 July), Multiverse Computing, which is based in San Sebastián, Spain, announced a $570m Series C funding round at a $1.7bn pre-money valuation. 

The funds are expected to go towards expanding the company’s model library, continuing research and development, investing in sovereign AI infrastructure and building out a regional presence in east and south-east Asia, the Middle East, Canada and the US.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

The Age Signals API will be available globally before the end of the year.

• 31.7% of surveyed employees admit to scrolling on their phones during meetings
• Distraction habits are taking over the workplace, with potentially devastating consequences for long-term employment prospects
• Directors are over 4 times more likely to arrive late for a meeting than an entry-level employee

Scrolling through your phone is now considered preferable to engaging in discussion in meetings, with almost a third (31.7%) of employees admitting to looking at their phone instead of the daily PowerPoint, new research has claimed.

Digital distractions in the workplace are costing employees of all ages opportunities to progress, while Gen Z employees are struggling with time management. The findings of a Careerminds UK survey of 600 UK employees highlight meetings, emails, and notifications delivering distraction from the focused task, with completion or attention to personal tasks while at work by 37.2% of respondents.

What just happened? Russia has charged Telegram founder and CEO Pavel Durov with “aiding terrorism” and placed him on an international wanted list. The Federal Security Service (FSB) claims the platform failed to remove channels, chats, and bots used by Ukrainian intelligence to organize attacks inside Russia. Telegram’s response was less than diplomatic: its official X account posted a photo of Durov giving the middle finger.

The charges follow a terrorism-related investigation that became public in February. The FSB alleges that Telegram was used to coordinate sabotage, terrorist attacks, mass killings, and cyberfraud, resulting in deaths and extensive property damage.

A large part of the case centers on Daivinchik/Leo, a popular Telegram dating chatbot in Russia, where Tinder is no longer available. According to the FSB and Russia’s Investigative Committee, Ukrainian agents posed as young women to contact and entrap young Russians before coercing them into committing crimes.

– Telegram Messenger (@telegram) July 29, 2026

Russian authorities say 46 people aged between 12 and 22 have been arrested over the last year after being recruited through the bot. They are accused of attacking law enforcement officers or setting fire to transport, energy, communications, and financial infrastructure. Ukraine has not commented on the allegations.

Durov, 41, could face life imprisonment if convicted. The FSB did not explain how it plans to add him to an international wanted list, and it’s unclear whether Moscow will seek an Interpol Red Notice. The billionaire lives in Dubai and holds French and Emirati citizenship. He posted that he was in Georgia last week, but his current location is unknown.

The case arrives amid Russia’s intensifying campaign against foreign internet services. Moscow has effectively throttled Telegram, forcing users to rely on VPNs, yet the Kremlin and Russian defense ministry continue posting on it daily. It was previously reported that the slowdown angered Russian soldiers and pro-war bloggers who depend on the app, while the Kremlin attempts to push citizens toward its state-backed Max service. A subsequent VPN crackdown was blamed for widespread banking outages.

Durov’s legal problems extend beyond Russia. French authorities arrested him in 2024 over allegations that Telegram failed to tackle criminal activity and cooperate with investigators. He denied wrongdoing and was eventually permitted to leave France while the investigation continued. Telegram later changed its terms to allow the disclosure of suspects’ IP addresses and phone numbers in response to valid legal orders.

Durov left Russia in 2014 after coming under government pressure over VKontakte, the social network he also founded. He has characterized Moscow’s actions against Telegram as an attempt to undermine privacy and free speech. The company’s response to the latest charge suggests that position has not changed.

Strands is the NYT’s latest word game after the likes of Wordle, Spelling Bee and Connections – and it’s great fun. It can be difficult, though, so read on for my Strands hints.

Want more word-based fun? Then check out my NYT Connections today and Quordle today pages for hints and answers for those games, and Marc’s Wordle today page for the original viral word game.

Enterprise technology has become significantly more interconnected over the past decade. Cloud platforms, automation, artificial intelligence, and software integrations have allowed organizations to operate with greater speed and flexibility, but they have also made understanding digital access far more complicated.

Every new application, automated workflow, and API introduces more identities with larger scope that require oversight, making visibility increasingly difficult as technology ecosystems continue to expand. A report shows that identity-based attacks accounted for nearly 1 in 3 intrusions observed in 2024, with attackers increasingly relying on valid accounts and stolen credentials as initial access methods.

Steve Emmanuel, CEO and founder of IntegraTRACE, witnessed that shift firsthand throughout more than a decade in software engineering. Beginning as a developer before advancing into technical leadership roles, he found that every step into larger organizations brought greater complexity. Teams expanded, cloud environments grew, and new applications were introduced, yet one challenge consistently remained.

“The biggest misconception is that somebody is keeping track of everything,” Emmanuel says. “Engineering assumes security has the answers. Security assumes engineering has the answers. Everyone believes someone has the complete picture, but very often there isn’t a reliable source of truth.”

Those experiences ultimately inspired the creation of IntegraTRACE, a SOC 2 Type II compliant identity visibility platform designed to help organizations understand how human, non-human, and agentic identities interact across cloud environments and enterprise applications. Rather than focusing solely on authentication or access management, Emmanuel explains that the company’s objective has been to provide organizations with a centralized view of digital identities as enterprise infrastructure becomes increasingly distributed.

From the beginning, Emmanuel wanted the platform to solve practical customer challenges rather than pursue features simply because they were popular. He explains that customer feedback has played a central role in product development, allowing the engineering team to respond quickly as cloud environments and identity ecosystems continue evolving.

According to Emmanuel, one of the most common responses from customers is realizing how much identity information had been dispersed across separate systems. Centralizing that information provides a clearer picture of access across the organization and helps surface potential risks before they become larger security concerns.

“Visibility changes the conversation,” Emmanuel says. “When organizations can see every identity and its level of access in one place, they gain insights that were previously difficult to obtain.”

From his perspective, customers frequently identify over-privileged accounts, dormant credentials, and permission structures that would have taken weeks or even months to uncover through manual reviews. Bringing those insights together into a centralized view allows security and engineering teams to spend less time gathering information and more time addressing potential risks.

IntegraTrace

He credits much of the company’s early growth to maintaining a close relationship with customers. Rather than following lengthy development roadmaps, the engineering team regularly incorporates customer feedback into the platform, often delivering requested improvements within weeks. Emmanuel believes that approach has helped ensure every product enhancement is tied to a practical business need.

“We’re building something people genuinely need,” he says. “When customers tell us there’s a problem slowing them down, we want to solve it quickly. That feedback continues to shape how the platform evolves.”

Looking ahead, Emmanuel says the company’s ambitions extend well beyond cloud infrastructure. Following early adoption across Canada and the United States, IntegraTRACE is preparing for expansion into Europe and Asia while broadening its visibility capabilities into HR systems, financial applications, and other enterprise software where identities continue to multiply.

His long-term vision is to make identity visibility a foundational layer across an organization’s technology ecosystem. As businesses continue adopting AI, automation, and increasingly connected applications, Emmanuel believes understanding digital access will become as important as managing the systems themselves.

“The organizations that adapt most successfully won’t necessarily be the ones with the most security tools,” Emmanuel says. “They’ll be the ones that have the clearest understanding of every identity operating across their business and the confidence to act on that knowledge.“

The KitchenAid KF3 Fully Automatic Espresso Machine is the middle child of the company’s 2026 release of a trio of compact bean-to-cup coffee machines, sitting below the KitchenAid KF4 and above the KF2. Because of its slim build, it’s a great option for smaller kitchens, but it can easily service a larger household thanks to four customizable user profiles and a solid selection of over 15 coffee-based drink recipes.

It benefits from KitchenAid’s IntelliGrind system — also featuring on the company’s larger machines like the KF8 and KF6 — that automatically determines the best grind size depending on the beans you pour into the top-mounted hopper. During my testing, I found it consistently delivered a well-extracted and tasty espresso.