Daily Brief

Former CIA analyst Asif William Rahman, 34, has been sentenced to 37 months in prison for mishandling classified information. Rahman unlawfully retained and transmitted top secret National Defense Information to unauthorized individuals. He was arrested in Cambodia, having worked for the CIA since 2016 with access to Sensitive Compartmented Information. Court documents reveal Rahman took classified documents home, photographed them, and sent the images to others after editing to conceal their origin. Some leaked documents, which detailed potential military actions between Israel and Iran, were shared on the platform Telegram. Rahman attempted to cover his tracks by deleting approximately 1.5 GB of data and modifying journal entries related to U.S. policies. His case highlights the swift action by U.S. authorities to uphold national security and deter similar acts of betrayal.

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has reported active exploitation of a Linux kernel flaw, CVE-2023-0386. This vulnerability, with a CVSS score of 7.8, allows for privilege escalation via an improper ownership bug in OverlayFS. CVE-2023-0386 was patched early in 2023, yet it continues to be a threat as attackers exploit systems that remain unpatched. The flaw enables unauthorized users to escalate privileges by manipulating file ownership in the Linux kernel. Datadog's analysis describes the vulnerability as trivial to exploit, involving the creation and execution of a root-owned SUID binary in directories like "/tmp". Additional related vulnerabilities, GameOver(lay) CVEs-2023-32629 and CVE-2023-2640, were identified, affecting Unix systems with similar privilege escalation risks. Federal Civilian Executive Branch (FCEB) agencies are mandated to patch this vulnerability by July 8, 2025, to safeguard against these exploits.

Veeam has released an update to fix a critical flaw (CVE-2025-23121) with a 9.9 CVSS score in its Backup & Replication software, allowing remote code execution. This vulnerability affects all prior builds of version 12 up to 12.3.1.1139, with the fix applied in version 12.3.2. The flaw was identified and reported by security researchers from CODE WHITE GmbH and watchTowr. A previous patch designed to address a similar vulnerability (CVE-2025-23120) was reportedly bypassable, highlighting ongoing security challenges. Another related vulnerability (CVE-2025-24286) addressed allows backup operators to potentially execute arbitrary code. Additional patches include fixes for Veeam Agent for Microsoft Windows, addressing a separate vulnerability (CVE-2025-24287) that could allow local users to execute code with elevated privileges. Rapid7 reveals that more than 20% of its incident responses in 2024 involved exploiting Veeam software once attackers were inside the network. It is critical for users to update their software promptly to mitigate risks associated with these vulnerabilities.

Iran has significantly throttled internet speeds following escalated conflicts, purportedly to block Israeli cyber attacks. Iranian spokesperson cited internet stability as the reason for this action, describing the slowdown as "temporary, targeted, and controlled." NetBlocks reported a marked decline in Iranian internet traffic on the day of the implementation. The cyber exchanges include attacks on Iran's Bank Sepah by a pro-Israel group, Predatory Sparrow, impacting the bank’s online and ATM services. Predatory Sparrow accuses Bank Sepah of supporting Iran's controversial activities, including its missile and nuclear programs. Heightened cyber activity from Iranian state actors and affiliated hacktivist groups has been observed, targeting various regional entities. The Iranian government has also advised citizens to delete WhatsApp, claiming, without evidence, that it is being used by Israel for espionage. The U.S. Department of State is seeking information on Iranian hackers suspected of targeting critical infrastructure in the U.S. and other countries using specific malware tools.

President Trump has decided to delay the enforcement of a law mandating TikTok to sell its U.S. operations or cease operations, marking this as the third delay. The law, titled "Protecting Americans from Foreign Adversary Controlled Applications Act," was initially passed with strong bipartisan support and aimed to safeguard U.S. national security. The deadline for TikTok's divestiture was initially set for January 19th, 2025, but has been extended multiple times, now moving to June 19th. Trump cited the ongoing U.S.-China trade disputes as a significant barrier to finalizing the sale and indicated that a resolution could expedite the process. Despite potential national security concerns, Trump acknowledged a personal affinity for TikTok, highlighting its role in his political campaigns. Currently, there is no resolution on the trade issues, nor clarity on potential American buyers for TikTok, leaving significant uncertainty about the app's future and security implications.

AWS has achieved 100% multi-factor authentication (MFA) enforcement for root users across all account types. Announcement made by AWS Chief Information Security Officer Amy Herzog at the re:Inforce cloud security conference. New security measures and capabilities rolled out at the conference to enhance cloud security for customers. AWS Identity and Access Management Access Analyzer introduced, providing insights into internal access to critical resources. AWS Security Hub now offers enhanced data analysis tools to help prioritize and respond to security issues more effectively. GuardDuty Extended Threat Detection, optimized for container-based applications, adds improved threat detection capabilities. AWS introduces network security director in Shield to manage network security posture and mitigate threats like DDoS and SQL injections. The implementation supports AWS's commitment to the CISA Secure by Design pledge, despite challenges like budget cuts and key staff departures.

Paddle.com and its U.S. subsidiary agreed to pay $5 million to settle FTC allegations of enabling deceptive tech support scams, primarily targeting U.S. consumers, including older adults. The UK-based payment processor was accused of inadequate screening and fraud prevention, which allowed foreign operators to exploit the U.S. credit card system with tech support scams. Scammers, including known offenders like Restoro, Reimage, and PC Vark, used fake virus alerts impersonating major brands to sell unnecessary software or services. Paddle processed significant amounts for these operators, over $12.5 million for PC Vark and $37 million for Restoro and Reimage, despite high complaint and chargeback rates. Internal communications at Paddle indicated awareness of the fraudulent activities and their impact on vulnerable consumers but took measures to hide these activities to evade bank and network scrutiny. The FTC claimed Paddle acted as an unregistered payment facilitator, breaching Visa and Mastercard rules, and continued relations with high-risk clients for profit. As part of the settlement, Paddle must adhere to strict conditions, including avoiding processing payments for entities engaged in deceptive practices. The case highlights the importance of consumer vigilance against unsolicited tech support claims and the necessity for businesses to maintain rigorous compliance and detection systems to prevent fraud.

TaxOff exploited a zero-day vulnerability (CVE-2025-2783) in Google Chrome to deploy the Trinper backdoor. The attack targeted Russian organizations, leveraging phishing emails designed as invitations to specific forums to distribute malicious links. Google patched this vulnerability after it was reported by Kaspersky during Operation ForumTroll. Trinper features include keystroke recording, file gathering, and remote command execution via a C2 server. Positive Technologies uncovered a similar past attack, also initiated via a phishing email disguised as a conference invitation, leading to speculation about connections to other hacking groups. Variations in the attack methodologies include the use of different malware loaders like Donut and Cobalt Strike. The backdoor employs multithreading technology enhancing concealment and parallel data processing. Strategic use of zero-day vulnerabilities and sophisticated malware indicates the group’s intent for long-term access and data exfiltration from targeted systems.

Scania confirmed a cybersecurity breach where attackers accessed insurance claim documents through stolen external IT partner credentials. The attackers threatened to leak the stolen data unless their ransom demands were met. The compromised system, "insurance.scania.com," was provided by an external IT partner. The breach was detected after threat actors posted on a hacking forum about selling the stolen data. Stolen documents might contain sensitive personal, financial, or medical data, heavily impacting affected individuals. Attackers used a ProtonMail address to send extortion emails directly to several Scania employees. Scania has notified privacy authorities and stated the breach had limited impact, but investigations are ongoing.

A high-risk vulnerability in LangChain's LangSmith platform, identified as AgentSmith, could allow attackers to steal sensitive data like API keys and user inputs. The flaw, carrying a CVSS score of 8.8, enabled unauthorized interception of communications between users and malicious agents configured with a proxy server. Attackers could deploy malicious agents on LangChain Hub, which users could then interact with, unknowingly routing data through the attackers' server. Compromised information may include OpenAI API keys and user prompts, leading to potential unauthorized model access, data breaches, and significant financial impacts. Following responsible disclosure and swift action by LangChain, the vulnerability was patched to prevent data interception and added warnings for potential proxy misuse. The security lapse had potential long-term effects, including legal liabilities and reputational harm for affected organizations due to unauthorized access to proprietary data and intellectual property. Separate reporting highlighted the emergence of new 'WormGPT' variants, demonstrating ongoing threats in the cybercrime arena linked to generative AI tools.

Security researchers at watchTowr uncovered multiple pre-authentication vulnerabilities in Sitecore CMS, prominently used by leading global companies. Three of the seven discovered vulnerabilities were publicly disclosed, allowing for full system control when chained together. One critical vulnerability involved hardcoded internal passwords, notably an internal account set with the password "b", which was brute-forced in just three seconds. Another vulnerability included a path traversal flaw in the ZIP unpacking mechanism, leading to remote code execution (RCE) when combined with the aforementioned password issue. A third disclosed vulnerability involved an unrestricted file upload flaw, also leading to RCE when exploited. Despite advisements against modifying default credentials, Sitecore's stance and documentation encouraged users to keep default settings, complicating security practices. These vulnerabilities were discovered in the most recent version of Sitecore at the time of the report, with patches only recently released for some issues. Over 22,000 Sitecore instances were found to be exposed, increasing the criticality of addressing these vulnerabilities in deployed environments.

Instagram ads fraudulently impersonating Canadian banks like BMO and EQ Bank have been targeting users with phishing scams and investment fraud. The ads misuse official bank branding and AI-generated deepfake videos to lure victims to phishing sites seeking personal and banking information. An example highlighted includes an ad for "Eq Marketing" that directs users to a counterfeit website, mimicking EQ Bank’s branding to collect banking credentials. Another scam ad features a fake "BMO Belski" story, misusing the name of BMO’s Chief Investment Strategist to attract potential investors to a fraudulent WhatsApp group. Scammers deploy ads via Facebook pages with follower counts likely bolstered by bots, utilizing the platform's pass-through capabilities to Instagram to avoid direct detection. Despite being reported, the fraudulent ads continued to appear on Instagram, prompting involvement from BMO and EQ Bank, and reaching out to Meta for further action. Meta has pledged to investigate and remove fraudulent content, while EQ Bank has already begun proactive measures on social media platforms to tackle these high-fidelity scams. These incidents highlight the importance of user vigilance on social media advertising and the necessity of verifying the authenticity of promotional claims through official channels.

Veeam issued security updates to address multiple vulnerabilities in their Backup & Replication software, including a critical RCE flaw. The RCE vulnerability, identified as CVE-2025-23121, affects systems integrated into domain environments and is exploitable by authenticated domain users. This flaw was specifically prevalent in Veeam Backup & Replication version 12 and later until it was resolved in the latest release, version 12.3.2.3617. Following insecure practices, many enterprises have integrated their backup servers into a Windows domain rather than following Veeam's advice to use separate Active Directory Forests and dual-factor authentication for admin accounts. Ransomware groups like Cuba and FIN7 have historically exploited similar Veeam vulnerabilities to facilitate data theft and hinder recovery by deleting backups prior to ransomware deployment. Veeam's backup solutions are widely used globally, including by 82% of Fortune 500 companies and 74% of the Global 2,000, making these vulnerabilities significant in scale and impact.

A series of vulnerabilities in Sitecore Experience Platform (XP) can lead to remote code execution (RCE) by exploiting a hardcoded password, "b," for an internal user account. The exploit chain involves three distinct vulnerabilities, starting with unauthorized authentication access through a backdoor login path. Attackers can gain an authenticated session by bypassing Sitecore's backend login checks, leading to subsequent exploitation steps. The second vulnerability exploited is a "Zip Slip" flaw within Sitecore’s Upload Wizard, allowing the execution of originally restricted paths. Potential attackers can upload malicious files, including web shells, directly into the webroot of the server, enabling further remote code execution. If the Sitecore PowerShell Extensions (SPE) module is installed, a third vulnerability allows arbitrary file uploads to any specified path, circumventing security controls. Over 22,000 Sitecore instances are publicly exposed, posing substantial security risks, especially to sectors like banking and global enterprises. Patches were released in May 2025 with detailed CVE publications delayed until June 17, 2025, urging immediate updates to mitigate potential exploitation risks.

Agentic AI systems function autonomously, adapting and acting with minimal human intervention, raising new cybersecurity challenges. These AI agents require complex identity and access management as they hold similar privileges within systems as human users do. Traditional identity and access management frameworks are inadequate for these dynamic, decision-making machines, necessitating new security models. Machine identities are being targeted by threat actors, exploiting AI's capabilities to mimic human identities and manipulate autonomous decisions. Organizations must adopt identity-first security strategies, treating AI agents as privileged entities and embedding strict controls and surveillance throughout the AI lifecycle. Delinea suggests a structured approach, including mapping AI agents’ access, aligning their privileges with specific tasks, and monitoring their activity with robust encryption and real-time logging. Swift action in redefining identity security frameworks and integrating advanced protections is essential to guard against evolving security threats and keep pace with technological advancements.