lookout.com - Massistant is a mobile forensics application used by law enforcement in China to collect extensive information from mobile devices.
Researchers at the Lookout Threat Lab have discovered a mobile forensics application named Massistant, used by law enforcement in China to collect extensive information from mobile devices. This application is believed to be the successor to a previously reported forensics tool named “MFSocket” used by state police and reported by various media outlets in 2019. These samples require physical access to the device to install, and were not distributed through the Google Play store.
Forensics tools are used by law enforcement personnel to collect sensitive data from a device confiscated by customs officials, at local or provincial border checkpoints or when stopped by law enforcement officers.
These tools can pose a risk to enterprise organizations with executives and employees that travel abroad - especially to countries with border patrol policies that allow them to confiscate mobile devices for a short period of time upon entry. In 2024, the Ministry of State Security introduced new legislation that would allow law enforcement personnel to collect and analyze devices without a warrant. There have been anecdotal reports of Chinese law enforcement collecting and analyzing the devices of business travellers. In some cases, researchers have discovered persistent, headless surveillance modules on devices confiscated and then returned by law enforcement such that mobile device activity can continue to be monitored even after the device has been returned.
forbes.com - Microsoft has confirmed that SharePoint Server is under mass attack and no patch is yet available — here’s what you need to know and how to mitigate the threat.
Microsoft Confirms CVE-2025-53770 SharePoint Server Attacks
It’s been quite the few weeks for security warnings, what with Amazon informing 220 million customers of Prime account attacks, and claims of a mass hack of Ring doorbells going viral. The first of those can be mitigated by basic security hygiene, and the latter appears to be a false alarm. The same cannot be said for CVE-2025-53770, a newly uncovered and confirmed attack against users of SharePoint Server which is currently undergoing mass exploitation on a global level, according to the Eye Research experts who discovered it. Microsoft, meanwhile, has admitted that not only is it “aware of active attacks” but, worryingly, “a patch is currently not available for this vulnerability.”
CVE-2025-53770, which is also being called ToolShell, is a critical vulnerability in on-premises SharePoint. The end result of which is the ability for attackers to gain access and control of said servers without authentication. If that sounds bad, it’s because it is. Very bad indeed.
“The risk is not theoretical,” the researchers warned, “attackers can execute code remotely, bypassing identity protections such as MFA or SSO.” Once they have, they can then “access all SharePoint content, system files, and configurations and move laterally across the Windows Domain.”
And then there’s the theft of cryptographic keys. That can enable an attacker to “impersonate users or services,” according to the report, “even after the server is patched.” So, even when a patch is eventually released, and I would expect an emergency update to arrive fairly quickly for this one, the problem isn’t solved. You will, it was explained, “need to rotate the secrets allowing all future tokens that can be created by the malicious actor to become invalid.”
And, of course, as SharePoint will often connect to other core services, including the likes of Outlook and Teams, oh and not forgetting OneDrive, the threat, if exploited, can and will lead to “data theft, password harvesting, and lateral movement across the network,” the researchers warned.
0din.ai - In a recent submission last year, researchers discovered a method to bypass AI guardrails designed to prevent sharing of sensitive or harmful information. The technique leverages the game mechanics of language models, such as GPT-4o and GPT-4o-mini, by framing the interaction as a harmless guessing game.
By cleverly obscuring details using HTML tags and positioning the request as part of the game’s conclusion, the AI inadvertently returned valid Windows product keys. This case underscores the challenges of reinforcing AI models against sophisticated social engineering and manipulation tactics.
Guardrails are protective measures implemented within AI models to prevent the processing or sharing of sensitive, harmful, or restricted information. These include serial numbers, security-related data, and other proprietary or confidential details. The aim is to ensure that language models do not provide or facilitate the exchange of dangerous or illegal content.
In this particular case, the intended guardrails are designed to block access to any licenses like Windows 10 product keys. However, the researcher manipulated the system in such a way that the AI inadvertently disclosed this sensitive information.
Tactic Details
The tactics used to bypass the guardrails were intricate and manipulative. By framing the interaction as a guessing game, the researcher exploited the AI’s logic flow to produce sensitive data:
Framing the Interaction as a Game
The researcher initiated the interaction by presenting the exchange as a guessing game. This trivialized the interaction, making it seem non-threatening or inconsequential. By introducing game mechanics, the AI was tricked into viewing the interaction through a playful, harmless lens, which masked the researcher's true intent.
Compelling Participation
The researcher set rules stating that the AI “must” participate and cannot lie. This coerced the AI into continuing the game and following user instructions as though they were part of the rules. The AI became obliged to fulfill the game’s conditions—even though those conditions were manipulated to bypass content restrictions.
The “I Give Up” Trigger
The most critical step in the attack was the phrase “I give up.” This acted as a trigger, compelling the AI to reveal the previously hidden information (i.e., a Windows 10 serial number). By framing it as the end of the game, the researcher manipulated the AI into thinking it was obligated to respond with the string of characters.
Why This Works
The success of this jailbreak can be traced to several factors:
Temporary Keys
The Windows product keys provided were a mix of home, pro, and enterprise keys. These are not unique keys but are commonly seen on public forums. Their familiarity may have contributed to the AI misjudging their sensitivity.
Guardrail Flaws
The system’s guardrails prevented direct requests for sensitive data but failed to account for obfuscation tactics—such as embedding sensitive phrases in HTML tags. This highlighted a critical weakness in the AI’s filtering mechanisms.
securityweek.com - The MITRE AADAPT framework provides documentation for identifying, investigating, and responding to weaknesses in digital asset payments.
The non-profit MITRE Corporation on Monday released Adversarial Actions in Digital Asset Payment Technologies (AADAPT), a cybersecurity framework designed to help the industry tackle weaknesses in cryptocurrency and other digital financial systems.
Modeled after the MITRE ATT&CK framework, AADAPT delivers a structured methodology that developers, financial organizations, and policymakers can use to find, investigate, and address risks in digital asset payments.
Insights that more than 150 sources from academia, government, and industry provided on real-world attacks on digital currencies and related technologies were used to create a playbook of adversarial TTPs linked to digital asset payment technologies.
The increased use of cryptocurrency has led to the emergence of sophisticated threats, such as phishing schemes, ransomware campaigns, and double-spending attacks, often with severe impact on organizations that lack cybersecurity resources, such as local governments and municipalities.
AADAPT is meant to help them enhance their stance through practical guidance and tools that specifically cover this financial market segment.
According to MITRE, AADAPT was founded on an in-depth review of underlying technologies such as smart contracts, distributed ledger technology (DLT) systems, consensus algorithms, and quantum computing, along with vulnerabilities and credible attack methods.
The tool supports critical use cases to help develop analytics for emulating threats, create detection techniques, compare insights, and assess security capabilities to prioritize decisions, essentially assisting stakeholders in adopting best practices.
“Digital payment assets like cryptocurrency are set to transform the future of global finance, but their security challenges cannot be ignored. With AADAPT, MITRE is empowering stakeholders to adopt robust security measures that not only safeguard their assets but also build trust across the ecosystem,” MITRE VP Wen Masters said.
next.ink - L'Hôpital privé de la Loire (HPL), qui se trouve à Saint-Étienne, a été victime d'une cyberattaque, révélait ce jeudi 10 juillet, le journal Le Progrès.
Géré par le groupe Ramsay, le HPL avait publié un communiqué de presse mardi 8 juillet affirmant qu'il avait été victime quelques jours plus tôt d'un « vol d’identité » concernant « une quantité importante de données personnelles de ses patients ». Mais l'établissement se voulait rassurant, affirmant que les données étaient « essentiellement de nature administrative ».
Mardi soir, une personne se présentant comme responsable du piratage a contacté nos confrères du Progrès pour s'en indigner. Elle affirme posséder des données concernant plus de 530 000 patients dont leurs cartes d'identité. Elle ajoute que « l'argent est la motivation » sans préciser le montant exigé.
Cette réaction a obligé l'hôpital à revoir sa communication. À l'AFP, il expliquait jeudi 10 juillet, avoir envoyé un email « à plus de 126 000 patients concernés par le piratage informatique de l’Hôpital privé de la Loire (HPL), et les 40 d’entre eux qui sont concernés par le vol de données médicales seront contactés individuellement ». Et il affirme que son fonctionnement n'a cependant pas été affecté.
Le parquet de Paris a, de son côté, expliqué à l'agence de presse que sa section cybercriminalité avait été saisie et avoir confié l'enquête à l’Office anticybercriminalité (OFAC).
Interrogé par l'AFP sur la demande de rançon, le groupe Ramsay n'a pas voulu s'exprimer sur le sujet.
msrc.microsoft.com - Microsoft is aware of active attacks targeting on-premises SharePoint Server customers. The attacks are exploiting a variant of CVE-2025-49706. This vulnerability has been assigned CVE-2025-53770.
SharePoint Online in Microsoft 365 is not impacted.
A patch is currently not available for this vulnerability. Mitigations and detections are provided below.
Our team is actively working to release a security update and will provide additional details as they are available.
How to protect your environment
To protect your on-premises SharePoint Server environment, we recommend customers configure AMSI integration in SharePoint and deploy Defender AV on all SharePoint servers. This will stop unauthenticated attackers from exploiting this vulnerability.
AMSI integration was enabled by default in the September 2023 security update for SharePoint Server 2016/2019 and the Version 23H2 feature update for SharePoint Server Subscription Edition. For more details on how to enable AMSI integration, see here.
If you cannot enable AMSI, we recommend you consider disconnecting your server from the internet until a security update is available.
We also recommend you deploy Defender for Endpoint to detect and block post-exploit activity.
We will continue to provide updates and additional guidance for our customers as they become available.
Microsoft Defender Detections and Protections
Microsoft Defender Antivirus
Microsoft Defender Antivirus provides detection and protection against components and behaviors related to this threat under the detection name:
Exploit:Script/SuspSignoutReq.A
Trojan:Win32/HijackSharePointServer.A
Microsoft Defender for Endpoint
Microsoft Defender for Endpoint provides customers with alerts that may indicate threat activity associated with this threat. These alerts, however, can be triggered by unrelated threat activity. The following alert titles in the Microsoft Defender Security Center portal can indicate threat activity on your network:
Possible web shell installation
Possible exploitation of SharePoint server vulnerabilities
Suspicious IIS worker process behavior
‘SuspSignoutReq’ malware was blocked on a SharePoint server
HijackSharePointServer’ malware was blocked on a SharePoint server
Advanced hunting
NOTE: The following sample queries let you search for a week’s worth of events. To explore up to 30 days’ worth of raw data to inspect events in your network and locate potential related indicators for more than a week, go to the Advanced Hunting page > Query tab, select the calendar dropdown menu to update your query to hunt for the Last 30 days.
To locate possible exploitation activity, run the following queries in Microsoft 365 security center.
Successful exploitation via file creation (requires Microsoft 365 Defender)
Look for the creation of spinstall0.aspx, which indicates successful post-exploitation of CVE-2025-53770. Run query in the Microsoft 365 Defender
DeviceFileEvents
| where FolderPath has "MICROS~1\WEBSER~1\16\TEMPLATE\LAYOUTS"
| where FileName =~ "spinstall0.aspx"
or FileName has "spinstall0"
| project Timestamp, DeviceName, InitiatingProcessFileName, InitiatingProcessCommandLine, FileName, FolderPath, ReportId, ActionType, SHA256
| order by Timestamp desc
ictjournal.ch - Pendant des années, le groupe de hackers pro-russe «Noname057(16)» a mené des attaques DDoS contre des serveurs occidentaux, y compris des infrastructures critiques en Suisse. Les autorités judiciaires ont désormais démantelé un botnet du groupe et procédé à des arrestations. Le Ministère public de la Confédération suisse (MPC) a émis trois mandats d’arrêt.
Les autorités judiciaires de plusieurs pays ont mené une opération coordonnée contre le groupe de hackers «Noname057(16)». Lors de l’Action-Day, lancée par Europol après plusieurs années d’enquête, des perquisitions ont eu lieu dans plusieurs pays, selon un communiqué du Ministère public de la Confédération suisse (MPC). Les autorités ont saisi des équipements et arrêté des personnes – tandis qu’en Suisse, «aucun ordinateur impliqué dans le réseau et dans les attaques ni aucune personne domiciliée dans le pays n’ont été identifiés».
Les mesures coordonnées à l’échelle internationale, baptisées Opération Eastwood, ont permis de démanteler un botnet constitué de plusieurs centaines de serveurs répartis dans le monde entier, selon l’Office fédéral de la police criminelle allemande (BKA). Le groupe «Noname057(16)» exploitait ce réseau pour lancer des attaques DDoS, des cyberattaques visant à surcharger délibérément des serveurs.
Trois mandats d’arrêt émis par la Suisse
Le groupe «Noname057(16)» s’est constitué un casier judiciaire conséquent ces dernières années. Le groupe pro-russe se manifeste régulièrement depuis le début de la guerre en Ukraine en mars 2022, indique le MPC. Ce collectif de hackers a mené des attaques DDoS contre de nombreux pays occidentaux qu’il considère comme pro-ukranien. À plusieurs reprises, des serveurs suisses, y compris des infrastructures sensibles, ont été ciblés. Ces attaques interviennent généralement lors d’événements liés à l’Ukraine.
Pour rappel, le groupe hacktiviste a paralysé les sites web du Parlement en été 2023, à l’occasion d’un discours vidéo du président ukrainien Volodymyr Zelensky devant l’Assemblée fédérale. En janvier 2024, les hackers sont redevenus actifs lors de la visite du président ukrainien au Forum économique mondial (WEF). Un an plus tard, les sites de la ville de Lucerne ainsi que de la Banque cantonale vaudoise ont également été ciblés. Des attaques hacktivistes ont aussi eu lieu en juin 2024 lors de la conférence de Bürgenstock pour la paix et pendant le Concours Eurovision de la chanson en mai 2025.
En juin 2023, le Ministère public de la Confédération a ouvert une enquête pénale contre des inconnus pour détérioration de données et contrainte, selon le communiqué. Dans le cadre des investigations internationales coordonnées, plusieurs membres du groupe de hackers ont pu être identifiés dont trois personnes clés présumées. Le MPC a étendu son enquête contre ces derniers et a émis des mandats d’arrêt à leur encontre.
Dans le cadre de l’Action-Day du 15 juillet 2025, les autorités de Suisse et d’Allemagne ont été rejointes par celles des États-Unis, des Pays-Bas, de la Suède, de la France, de l’Espagne et de l’Italie. L’opération a bénéficié du soutien d’Europol, d’Eurojust et d’autres pays européens, précise la police fédérale allemande (BKA). En Suisse, le MPC et l'Office fédéral de la police (Fedpol) ont contribué à l'enquête.
Le MPC considère les résultats de l’opération comme la preuve que «les autorités de poursuite pénale sont aussi en mesure d’identifier des cybercriminels hautement professionnels et d’offrir une protection contre leurs attaques». Le MPC souligne l’importance de la coopération internationale dans la lutte contre la cybercriminalité transfrontalière.
theregister.com - Exclusive Aviation insiders say Serbia's national airline, Air Serbia, was forced to delay issuing payslips to staff as a result of a cyberattack it is battling.
Internal memos, seen by The Register, dated July 10 told staff: "Given the current situation and the ongoing cyberattacks, for security reasons, we will postpone the distribution of the June 2025 payslips.
"The IT department is working to resolve the issue as a priority, and once the conditions allow, the payslips will be sent to your email addresses."
Staff were reportedly paid their monthly salaries, but access to their payslip PDF was unavailable.
HR warned staff earlier in the day against opening emails that appeared to be related to payslips, or those that mention the staff members' first and last names "as if you sent them to yourself."
"We also kindly ask that you act responsibly given the current situation."
According to other internal comms seen by The Register, Air Serbia's IT team began emailing staff warning them that it was facing a cyberattack on July 4.
"Our company is currently facing cyberattacks, which may lead to temporary disruptions in business processes," they read.
"We kindly ask all managers to promptly create a work plan adapted to the changed circumstances, in accordance with the Business Continuity Plan, and to communicate it to their teams as soon as possible."
The same email communication chain mentioned the company's IT and security manager issuing a staff-wide password reset and installing security-scanning software on their machines on July 7.
All service accounts were killed at this point, which affected several automated processes, and datacenters were added to a demilitarized zone, which led to issues with users not being able to sync their passwords.
Additionally, internet access was removed for all endpoints, leaving only a certain few whitelisted pages under the airserbia.com domain available.
IT also installed a new VPN client "due to identified security vulnerabilities."
"We kindly ask you to take this situation seriously and fully cooperate with the IT team," the memo reads. "Please allow them to install the necessary software as efficiently as possible and carefully follow any further instructions they provide."
Two days after this, another wave of password resets came, the source said. Instead of allowing users to choose their own, the replacements followed a template from the sysadmins.
On July 11, IT issued a third wave of password resets, and staff were asked to leave their PCs locked but open before heading home for the weekend, so the IT team could continue working on them.
A source familiar with the matter, who spoke to The Register on condition of anonymity, said Air Serbia is trying to clean up a cyberattack that led to a deep compromise of its Active Directory.
As of July 14, the source claimed the airline's blue team has not fully eradicated the attackers' access to the company network and is not sure when the attackers broke in, due to a lack of security logs, although it is thought to be in the first few days of July.
The attack at the company, which is government-owned, is likely to have led to personal data compromise, the insider suspects, and some staff expressed concern that the company might not publicly disclose the intrusion.
bankinfosecurity.com - Hacker Claims to Have Exploited Flaw in Oracle WebLogic Server, Sold Stolen Data
A hacker claims to have stolen and sold the personal data of clients of Seychelles Commercial Bank. The bank, which provides personal and corporate services on Seychelles, one of the world's smallest countries, notified customers of a hack, but said only personal information - not money - was stolen.
The archipelago nation in the Indian Ocean, located northeast of Madagascar, sports 98,000 inhabitants, ranks as the richest country in Africa and has a reputation for being a tax haven.
Seychelles Commercial Bank on Friday said it "recently identified and contained a cybersecurity incident, which has resulted in its internet banking services being temporarily suspended," and requested customers "make use of our ATMs or visit one of our branches during normal banking hours."
In its breach notification, the bank told customers: "SCB regrets to inform that this cyber incident resulted in unintentional exposure of personal information of internet banking customers only. The bank reassures all its internet banking customers that no funds have been accessed."
Cyberattack disrupted UNFI’s operations in June; company estimates $50–$60 million net income hit but anticipates insurance will cover most losses.
United Natural Foods, Inc. (NYSE: UNFI), the main distributor for Amazon’s Whole Foods, said the June 2025 cyberattack that caused disruptions to business operations will impact fiscal 2025 net sales by an estimated $350 to $400 million.
In an update on Wednesday, the Rhode Island-based natural food products giant said “anticipated insurance” proceeds would significantly offset those loses.
“The Company estimates that the cyber incident will impact fiscal 2025 net sales by approximately $350 to $400 million, net (loss) income by $50 to $60 million, which includes the estimated tax impact, and adjusted EBITDA by approximately $40 to $50 million,” the company said in a business update on July 16th. “These estimates do not reflect the benefit of anticipated insurance proceeds, which the Company expects will be adequate for the incident. The Company does not currently expect a meaningful operational or financial impact beyond the fourth quarter of fiscal 2025 aside from insurance reimbursement.”
The company revealed in a filing with the SEC on June 9 that it had detected unauthorized activity on some IT systems on June 5. In response to the intrusion, certain systems were taken offline, which impacted its ability to fulfill and distribute customer orders.
UNFI advertises itself as the largest full-service grocery partner in North America, delivering products to over 30,000 locations, including natural product superstores, conventional supermarket chains, e-commerce providers, and independent retailers. With more than $30 billion in annual revenue, the company offers more than 250,000 natural, organic and conventional SKUs through its more than 50 distribution centers.
“We are grateful to our customers, suppliers, and associates for their resilience and collaboration as we worked through a challenging period for all of us. With our operations returning to more normalized levels, we remain focused on adding value for our customers and suppliers while becoming a more efficient and effective partner,” said Sandy Douglas, UNFI’s CEO.
The Company updated its full-year outlook to reflect its strong performance for the first three fiscal quarters of 2025 and the estimated costs and charges associated with the June cyber incident.
koreaherald.com - Seoul Guarantee Insurance, South Korea's largest provider of guarantee insurance, has been crippled by a ransomware attack, with its core systems offline for a third straight day.
The incident began early Monday, when SGI reported an “abnormal symptom” in its database system. By Tuesday afternoon, a joint investigation by the Financial Supervisory Service and the Financial Security Institute confirmed it was caused by a ransomware breach.
As a pivotal player in Korea’s guarantee insurance industry, SGI’s disruption is generating widespread confusion and inconvenience. The insurer provides guarantees for both individuals and corporations, with a guarantee balance of 478 trillion won ($344.4 billion) as of end-2024.
The impact is particularly severe in the housing market, where many rely on guarantee insurance for the “jeonse” rental system, where renters pay a large, refundable deposit in exchange for no monthly rent. SGI is one of the leading providers in this space, offering the highest cap on jeonse loan guarantees at 500 million won, compared to 200 million to 400 million won from other institutions.
While some services have been restored through cooperation with financial institutions, SGI’s main data system remains inoperative as of Wednesday morning. In urgent cases, the company has resorted to issuing handwritten guarantee certificates to minimize disruption.
Starting Wednesday, the insurer is operating an emergency center to collect reports of consumer damage and support recovery. “We vow full compensation and are planning responsible follow-up measures,” said SGI President and CEO Lee Myung-soon.
This is the first full-system disruption at a Korean financial institution caused by a ransomware attack and a second such case involving a Korean company this year. In June, major online bookstore Yes24 experienced a five-day outage and an estimated 10 billion won in lost sales due to a similar breach.
mobile-hacker.com - On June 13, 2025 was disclosed vulnerability in the iOS version of the Air Keyboard app that exposes users to remote input injection over Wi-Fi. The flaw, documented in CXSecurity Report, allows an attacker on the same local network to send keystrokes to a target iOS device without authentication. As of this writing, the app remains available on the App Store and is still affected by the vulnerability. With the report is also published prove of concept python script. In this blog I will test the exploit, have a look on their Android version of Air Keyboard app and conclude with security tips.
According to its official information, Air Keyboard is an app that turns your mobile device into a wireless keyboard and mouse for your computer. It connects over the local network and sends or receives input to or from a companion desktop application installed on Windows or macOS. The app’s goal is to offer convenient remote control for presentations, media playback, or general PC use, all from your smartphone or tablet.
The vulnerability stems from the iOS app listening on TCP port 8888 for incoming input — without any form of authentication or encryption. A proof-of-concept Python script included in the advisory demonstrates how an attacker can craft data and remotely inject arbitrary keystrokes to the victim’s device. A video demonstration further confirms how trivial the attack is to execute. Because the iOS app does not verify the origin or integrity of the incoming commands, any device on the same Wi-Fi network can send input as if it were the user.
The app remains available on the App Store in this vulnerable state, with no fix or warning issued to users.
europol.europa.eu - Between 14 and 17 July, a joint international operation, known as Eastwood and coordinated by Europol and Eurojust, targeted the cybercrime network NoName057(16). Law enforcement and judicial authorities from Czechia, France, Finland, Germany, Italy, Lithuania, Poland, Spain, Sweden, Switzerland, the Netherlands and the United States took simultaneous actions against offenders and infrastructure belonging to the pro-Russian cybercrime network. The investigation was also supported by ENISA, as well as Belgium, Canada, Estonia, Denmark, Latvia, Romania and Ukraine. The private parties ShadowServer and abuse.ch also assisted in the technical part of the operation.
The actions led to the disruption of an attack-infrastructure consisting of over one hundred computer systems worldwide, while a major part of the group's central server infrastructure was taken offline. Germany issued six warrants for the arrest of offenders living in the Russian Federation. Two of these persons are accused of being the main instigators responsible for the activities of "NoName057(16)". In total, national authorities have issued seven arrest warrants, which are directed, inter alia, against six Russian nationals for their involvement in the NoName057(16) criminal activities. All of the suspects are listed as internationally wanted, and in some cases, their identities are published in media. Five profiles were also published on the EU Most Wanted website.
National authorities have reached out to several hundred of individuals believed to be supporters of the cybercrime network. The messages, shared via a popular messaging application, inform the recipient of the official measures highlighting the criminal liability they bear for their actions pursuant to national legislations. Individuals acting for NoName057(16) are mainly Russian-speaking sympathisers who use automated tools to carry out distributed denial-of-service (DDoS) attacks. Operating without formal leadership or sophisticated technical skills, they are motivated by ideology and rewards.
Learn how the JFrog Security research team discovered and disclosed CVE-6514, a critical vulnerability in the mcp-remote project used by Model Context Protocol clients.
WASHINGTON, July 15 (Reuters) - A U.S. state's Army National Guard network was thoroughly hacked by a Chinese cyberespionage group nicknamed "Salt Typhoon," according to a Department of Homeland Security memo.
The memo obtained by Property of the People, a national security transparency nonprofit, said the hackers "extensively compromised" the unnamed state Army National Guard's network between March and December 2024 and exfiltrated maps and "data traffic" with counterparts' networks in "every other US state and at least four US territories."
he National Guard and the Department of Homeland Security's cyber defense arm, CISA, did not immediately return messages. News of the memo was first reported by NBC News.
Salt Typhoon has emerged as one of the top concerns of American cyber defhen Coatesenders. U.S. officials allege that the hacking group is doing more than just gathering intelligence; it is prepositioning itself to paralyze U.S. critical infrastructure in case of a conflict with China. Beijing has repeatedly denied being behind the intrusions.
The memo, which said it drew on reporting from the Pentagon, said that Salt Typhoon's success in compromising states' Army National Guard networks nationwide "could undermine local cybersecurity efforts to protect critical infrastructure," in part because such units are often "integrated with state fusion centers responsible for sharing threat information—including cyber threats."
We tested Grok 4 – Elon’s latest AI model – and it failed key safety checks. Here’s how SplxAI hardened it for enterprise use.
On July 9th 2025, xAI released Grok 4 as its new flagship language model. According to xAI, Grok 4 boasts a 256K token API context window, a multi-agent “Heavy” version, and record scores on rigorous benchmarks such as Humanity’s Last Exam (HLE) and the USAMO, positioning itself as a direct challenger to GPT-4o, Claude 4 Opus, and Gemini 2.5 Pro. So, the SplxAI Research Team put Grok 4 to the test against GPT-4o.
Grok 4’s recent antisemitic meltdown on X shows why every organization that embeds a large-language model (LLM) needs a standing red-team program. These models should never be used without rigorous evaluation of their safety and misuse risks—that's precisely what our research aims to demonstrate.
Key Findings
For this research, we used the SplxAI Platform to conduct more than 1,000 distinct attack scenarios across various categories. The SplxAI Research Team found:
With no system prompt, Grok 4 leaked restricted data and obeyed hostile instructions in over 99% of prompt injection attempts.
With no system prompt, Grok 4 flunked core security and safety tests. It scored .3% on our security rubric versus GPT-4o's 33.78%. On our safety rubric, it scored .42% versus GPT-4o's 18.04%.
GPT-4o, while far from perfect, keeps a basic grip on security- and safety-critical behavior, whereas Grok 4 shows significant lapses. In practice, this means a simple, single-sentence user message can pull Grok into disallowed territory with no resistance at all – a serious concern for any enterprise that must answer to compliance teams, regulators, and customers.
This indicates that Grok 4 is not suitable for enterprise usage with no system prompt in place. It was remarkably easy to jailbreak and generated harmful content with very descriptive, detailed responses.
However, Grok 4 can reach near-perfect scores once a hardened system prompt is applied. With a basic system prompt, security jumped to 90.74% and safety to 98.81%, but business alignment still broke under pressure with a score of 86.18%. With SplxAI’s automated hardening layer added, it scored 93.6% on security, 100% on safety, and 98.2% on business alignment – making it fully enterprise-ready.
securityweek.com - DragonForce says it stole more than 150 gigabytes of data from US department store chain Belk in a May cyberattack
The DragonForce ransomware gang has claimed responsibility for a disruptive cyberattack on US department store chain Belk.
The incident was identified on May 8 and prompted Belk to disconnect affected systems, restrict network access, reset passwords, and rebuild impacted systems, which disrupted the chain’s online and physical operations for several days. The company’s online store is still offline at the time of publication.
Belk’s investigation into the attack determined that hackers had access to its network between May 7 and May 11, and that they exfiltrated certain documents, including files containing personal information.
In a data breach notification submitted to the New Hampshire Attorney General’s Office, Belk said at least names and Social Security numbers were compromised in the attack.
The company is providing the impacted individuals with 12 months of free credit monitoring and identity restoration services, which also include up to $1 million identity theft insurance.
The company has not named the group responsible for the attack, but the DragonForce ransomware gang has claimed the incident on Monday, adding Belk to its Tor-based leak site.
theguardian.com - Conservative government used superinjuction to hide error that put Afghans at risk and led to £2bn mitigation scheme.
Thousands of Afghans relocated to UK under secret scheme after data leak
Conservative government used superinjuction to hide error that put Afghans at risk and led to £2bn mitigation scheme
What we know about the secret Afghan relocation scheme
Afghan nationals: have you arrived in the UK under the Afghan Response Route?
Dan Sabbagh and Emine Sinmaz
Tue 15 Jul 2025 22.07 CEST
Share
Conservative ministers used an unprecedented superinjunction to suppress a data breach that led the UK government to offer relocation to 15,000 Afghans in a secret scheme with a potential cost of more than £2bn.
The Afghan Response Route (ARR) was created in haste after it emerged that personal information about 18,700 Afghans who had applied to come to the UK had been leaked in error by a British defence official in early 2022.
Panicked ministers and officials at the Ministry of Defence learned of the breach in August 2023 after data was posted to a Facebook group and applied to the high court for an injunction, the first sought by a British government – to prevent any further media disclosure.
It was feared that publicity could put the lives of many thousands of Afghans at risk if the Taliban, who had control of the country after the western withdrawal in August 2021, were to become aware of the existence of the leaked list and to obtain it.
The judge in the initial trial, Mr Justice Knowles, granted the application “contra mundum” – against the world – and ruled that its existence remain secret, resulting in a superinjunction which remained in place until lifted on Tuesday.
The gagging order meant that both the data breach and the expensive mitigation scheme remained hidden despite its size and cost until the near two-year legal battle was brought to a close in the high court.
At noon on Tuesday, the high court judge Mr Justice Chamberlain said it was time to end the superinjuction, which he said had the effect of concealing discussions about spending “the sort of money which makes a material difference to government spending plans and is normally the stuff of political debate”.
A few minutes later, John Healey, the defence secretary, offered a “sincere apology” for the data breach. In a statement to the Commons, he said he had felt “deeply concerned about the lack of transparency” around the data breach and “deeply uncomfortable to be constrained from reporting to this house”.
propublica.org - The Pentagon bans foreign citizens from accessing highly sensitive data, but Microsoft bypasses this by using engineers in China and elsewhere to remotely instruct American “escorts” who may lack expertise to identify malicious code.
Microsoft is using engineers in China to help maintain the Defense Department’s computer systems — with minimal supervision by U.S. personnel — leaving some of the nation’s most sensitive data vulnerable to hacking from its leading cyber adversary, a ProPublica investigation has found.
The arrangement, which was critical to Microsoft winning the federal government’s cloud computing business a decade ago, relies on U.S. citizens with security clearances to oversee the work and serve as a barrier against espionage and sabotage.
But these workers, known as “digital escorts,” often lack the technical expertise to police foreign engineers with far more advanced skills, ProPublica found. Some are former military personnel with little coding experience who are paid barely more than minimum wage for the work.
