The Register

Chipzilla hopes agents, robots, and edge devices make CPUs cool again... now it has to build the chips

Intel is betting on AI to reverse its fortunes, wagering that inference and agentic workloads will restore the CPU to the center of compute - even as its chip manufacturing struggles persist.…

Ailing scaling blamed by Windows-maker for unreadable missives

Microsoft's update to harden Remote Desktop against phishing attacks has arrived. When users open a Remote Desktop (.rdp) file, they should now see a warning listing all requested connection settings - or they would if it was displaying correctly.…

BLACK HAT ASIA Open source models can find bugs as effectively as Anthropic's Mythos, according to Ari Herbert-Voss, CEO of AI-powered security startup RunSybil and OpenAI's first security hire. Speaking at the Black Hat Asia conference in Singapore today, Herbert-Voss said Mythos excels at finding both "shallow" bugs - well-described flaws that are and easy to validate - and more complex vulnerabilities. In his talk, he attributed this to "supralinear scaling": where researchers assumed LLM capability would improve linearly, evidence now suggests a model trained on twice the data, compute, and time produces something four times more capable. He hinted supralinear scaling might produce even better multipliers but could not say more due to a non-disclosure agreement. Anthropic has kept access to Mythos tghtly restricted, citing fears of misuse. However Herbert-Voss argues attackers and defenders alike can achieve comparable results with open source models by building "scaffolding" to run several of them in harness. That approach also improves defense in depth, as different models tend to catch different flaws — a useful hedge against any single model's blind spots. Cost is another driver. Mythos is expensive to build and run, and may never be publicly available, making open source alternatives not just viable but necessary for many organizations. Herbert-Voss feels human expertise is still needed to orchestrate open source models so they together deliver Mythos-grade performance, and to assess the bug reports AI generates. He then noted that fuzzing, the testing technique which injects random or near-random data into software to see if doing so produces bugs, also creates so many warnings that it can make extra work for humans. AI bug-hunters already produce the same problem, and he expects it will persist. Herbert-Voss therefore thinks infosec workers will have plenty on their plates for the foreseeable future, and the economic incentive to use AI – someone's got to use services that pay for all those GPUs and datacenters – will act as a forcing function that makes infosec teams adopt AI and as a result improve their proactive and defensive work. ®

OpenAI's first security hire, Ari Herbert-Voss, thinks more automated bug finding will improve security without costing jobs

Black Hat Asia  Open source models can find bugs as effectively as Anthropic's Mythos, according to Ari Herbert-Voss, CEO of AI-powered security startup RunSybil and OpenAI's first security hire.…

Greece is taking a flexible approach to introducing the European Union's biometric Entry/Exit System (EES), after some British passport holders missed flights home following the system's implementation on 10 April. On 20 April, the Hellenic Police said it had started to fully deploy EES and Greece will use all necessary measures to manage traffic at border crossings in accordance with the regulations. The UK government updated its travel advice on the same day: "Greek authorities have indicated that they will not collect biometric data (fingerprints and photos) for UK travellers as part of EES" although passengers should "follow the advice of authorities on the ground". The Independent's travel correspondent Simon Calder said EES regulations allow flexibility to suspend biometric data collection at specific places for limited periods of time, yet the European Commission does not expect blanket exemptions for citizens of specific countries for extended periods. EES went live in October last year, causing launch-day queues at Prague's international airport. Many countries phased it in, but its use became compulsory on 10 April. Since then some Brits have missed flights due to border control delays caused by the system, including around 100 passengers due to take an Easyjet flight from Milan Linate to Manchester on 12 April and a smaller number trying to travel with Ryanair from Milan Bergamo to Manchester on 16 April. "The issues some passengers have experienced are unacceptable and so we continue to urge border authorities to ensure they make full and effective use of the permitted flexibilities for as long as needed, so our customers' travel plans are not impacted," Easyjet said in a statement. The airline will transfer passengers delayed by EES to other flights it provides for free, something it offered to a Register staffer and his family when they were delayed trying to fly from Paris to London Gatwick on 10 April. As replacement flights were not available until 15 days later, our vulture decided to catch a Eurostar instead. Airport operator group ACI Europe and Airlines for Europe, which includes Easyjet, Ryanair and British Airways parent company International Airlines Group, said some passengers have waited two to three hours due to the system. "Border control authorities must be allowed to fully suspend the EES when waiting times become excessive," said the heads of the two organizations. "This is essential not only in the coming weeks, but throughout the peak summer travel season." As of 10 April, the European Commission said that EES had registered more than 52 million entries and exits, refused entry to more than 27,000 individuals (about 0.1 percent of those trying to enter) and identified more than 700 as security risks. The EU has built a Travel to Europe app allowing those required to use EES to undertake some of the process in advance. However, so far only Sweden and Portugal are accepting its use. ®

Missed flights and more means something has got to give at the border

Greece is taking a flexible approach to introducing the European Union's biometric Entry/Exit System (EES), after some British passport holders missed flights home following the system's implementation on 10 April.…

Nothing says 'We want honest opinions' like a 36,000-letter mailshot with no awkward questions allowed

Members of the UK government’s People’s Panel on Digital ID will spend two weekends in Birmingham and three evenings on Zoom discussing how Britain should build a national digital identity system, earning £550 plus expenses for their trouble.…

BLACK HAT ASIA Infosec outfit SentinelOne found malware that tries to induce errors in engineering and physics simulation software and therefore represents an attempt at sabotage, and suggests it was created years before the Stuxnet worm that aimed to destroy Iran’s uranium enrichment centrifuges. The company’s Vitaly Kamluk discussed the malware in a talk at the Black Hat Asia conference today. SentinelOne has also published a blog post about the malware. Kamluk told the conference the discovery came about after he wondered if known nation-state-espionage tools like Flame, Animal Farm, and Project Sauron were the first of their kind. All three shared use of the Lua language and virtual machine, so he went looking for similar software. That search led to a malware sample uploaded to VirusTotal in 2016 that includes a reference to “fast16”. Kamluk’s analysis of the sample suggested the techniques its developers employed were not typical of 2016-era malware. SentinelOne researchers also recalled that the infamous ShadowBroker malware trove that appeared in 2016 and which was later linked to the United States National Security Agency, contained a reference to fast16. SentinelOne thinks fast16 came into existence around 2005, based on clues in the code and the fact it won’t run on anything more recent than Windows XP – and even then only on a single-core CPU. Intel shipped its first multi-core consumer CPUs in 2006. The researchers analyzed the sample and found it tries to install a worm and deploy a driver called fast16.sys. The driver includes a routine that alters the output of floating-point calculations and also goes looking for “precision calculation tools in specialised domains such as civil engineering, physics and physical process simulations.” The researchers think fast16 targeted three high-precision engineering and simulation suites that were used in the mid-2000s: “LS-DYNA 970, PKPM, and the MOHID hydrodynamic modeling platform, all used for scenarios like crash testing, structural analysis, and environmental modeling.” Iran is thought to have used LS-DYNA in its nuclear weapons program. Kamluk hypothesized that fast16’s purpose was to cause errors in calculations run by engineering simulation software, perhaps leading to real-world problems. And he asserted that fast16 was a cyberweapon that preceded Stuxnet by five years. “In the broader picture of APT evolution, fast16 bridges the gap between early, largely invisible development programs and later, more widely documented Lua‑ and LuaJIT‑based toolkits,” Kamluk wrote with SentinelOne colleague Juan Andrés Guerrero-Saade. “It is a reference point for understanding how advanced actors think about long‑term implants, sabotage, and a state’s ability to reshape the physical world through software. fast16 was the silent harbinger of a new form of statecraft, successful in its covertness until today.” In his talk, Kamluk said he’s disclosed his work to the vendors of the engineering applications fast16 targets, because he feels they may want to check the output of their products for evidence that the malware produced incorrect calculations. “Maybe there are more discoveries to come?” he concluded. Kamluk tearfully dedicated his talk to friend and colleague Sergey Mineev, who he said was responsible for finding many enormously significant APTs, without seeking attention for the significance of his work, and passed away in March. ®

FAST16 could be the first cyberweapon, and its effects could be with us today

Black Hat Asia  Infosec outfit SentinelOne found malware that tries to induce errors in engineering and physics simulation software and therefore represents an attempt at sabotage, and suggests it was created years before the Stuxnet worm that aimed to destroy Iran’s uranium enrichment centrifuges.…

BLACK HAT ASIA Developers of rented internet of things infrastructure – stuff like public EV chargers and shared e-bikes – are prioritizing user convenience over security, and leaving themselves exposed to wide-scale denial of service attacks on their services. That frightening thesis was the subject of a Friday talk at the Black Hat Asia conference, delivered by Hetian Shi, a hardware and IoT security researcher at China’s Tsinghua University. Shi told the conference the very nature of rented IoT services means they have a unique security problem: Anyone can access devices and examine them for vulnerabilities. The researcher conducted his probes with permission, and disclosed the results ethically – for which we should all be thankful because he discovered that some rentable devices include either a debugging port or a UART connector that makes examining their operations an uncomplicated task for an educated attacker. His own efforts yielded evidence of shared authentication keys in device firmware, and backend services that don’t properly authenticate users. The researcher also investigated the apps that rentable IoT providers publish so consumers can access their services and again found weak security that allowed him to do things like create phantom clients that rentable IoT services could not distinguish from actual customers. Using phantom clients makes it possible for an attacker to charge cars or rent scooters at zero cost. Shi said the techniques he’s developed can also compromise personal information by exposing rentable IoT services’ back ends. He’s created a tool called “IDScope” that makes it possible to exploit many of the flaws he found and during his talk demonstrated it by running the iOS app for a Chinese provider of public electric vehicle charging stations. Shi asked the audience to nominate a Chinese city – Shanghai was the popular choice – and then looked up available chargers in People’s Square, a major shopping and recreation district. The app produced a list of chargers and which ones were available to use. Shi asked the audience to choose which of the available chargers he should attack, noted the ID number for that charger listed in the app, entered that number into a script. A second or two later, the icon in the app for that charger changed color from green – which denotes availability for charging – to the grey hue that indicates a disabled port. The app was in Chinese and your correspondent can’t read that language so I can’t say with certainty what I witnessed, but the demo drew spontaneous applause from others in the audience – and plenty of people here at Black Hat have come from the Chinese-speaking world. Shi thinks the techniques he created also make it possible to deny service, and do so at scale – creating the possibility of taking out an entire city’s network of EV chargers. And not just in China: The researcher tested 11 apps published by European providers of shared bikes and scooters, and found similar problems - suggesting his findings will be applicable elsewhere. He theorized that the flaws he found are the result of developers trying to build services that users find convenient, at the expense of security. ®

Demonstrated in China, probably applicable elsewhere

Black Hat Asia  Developers of rented internet of things infrastructure – stuff like public EV chargers and shared e-bikes – are prioritizing user convenience over security, and leaving themselves exposed to wide-scale denial of service attacks on their services.…

Legit-looking website, camera-on interviews, jokes about backdoors ... it worked

EXCLUSIVE  It all started with a LinkedIn message, as so many employment scams do these days.…

All the Typhoons, everywhere, all at once

A majority of China-linked threat actors are using compromised routers and IoT devices worldwide, turning this gear into proxy networks to carry out further intrusions, steal sensitive data, and disrupt victim organizations’ operations, according to a joint 10-country advisory.…

Push to protect minors risks hitting everyone online

Proton's boss has waded into the age verification fight with a warning that sounds less like child safety and more like an identity checkpoint for the entire internet.…

Wins $300M deal over Salesforce, IBM because of 'integration with existing USDA systems,' among other things

Palantir has won a $300 million contract from the US Department of Agriculture (USDA) to support the National Farm Security Action Plan (NFSAP) and modernize how USDA delivers services to America's farmers.…

World's largest biomedical dataset lifted and shifted on Chinese mega marketplace

Breaking  Details of volunteers of UK-based Biobank, which describes itself as the custodian of the world's most comprehensive biomedical dataset, are for sale on Chinese ecommerce site Alibaba.…

Windows Admin Center flaws mean on-prem can attack cloud, and vice-versa

Black Hat Asia  Israeli researchers found a series of flaws in Microsoft's Windows Admin Center (WAC) and suggest this shows hybrid cloud management tools are a two-way attack surface that users don't spend enough time worrying about.…

Orgs can now buy UK cyber agency engineered commercial gadget, but details are slim

GCHQ's cyber arm has entered the hardware game with its first device designed to prevent cyberattacks on display devices.…

Keeping it simple for the developers can lead to very complex headaches later

PWNED  Welcome back to PWNED, the column where we celebrate the people who’ve taught us how not to secure a server. If you’ve ever tied your own shoelaces together, then tripped over them, or attempted to dive into a swimming pool but hit your head on the diving board, we’ll be talking about your cyber equivalent.…

NCSC passes judgment: passkeys pass muster, passwords fail

The UK's National Cyber Security Centre (NCSC) has officially endorsed passkeys as the default authentication standard, marking the first time the agency has told consumers to move away from passwords entirely.…