Full Daily Digest

The exploitation of CVE-2026-35273 highlights a sophisticated attack chain involving unauthenticated RCE, custom MeshCentral agents masquerading as legitimate cloud endpoints, and lateral movement scripts like [victim_abbreviation]_fanout.sh. ShinyHunters' focus on education suggests deliberate targeting of high-value data. The absence of patches forces reliance on mitigations, increasing risk. Organizations must prioritize hardening PeopleSoft instances, monitoring for IoCs, and segmenting critical systems. The campaign's correlation with data leaks on ShinyHunters' DLS underscores the urgency of response. Google's remediation guidance, including network segmentation and endpoint monitoring, is critical for defense.

SecurityWeek · 2026-06-12 · Incidents: ShinyHunters exploited a critical Oracle PeopleSoft zero-day (CVE-2026-35273) to steal data, primarily targeting the education sector.

The App.MenuItem artifact in macOS Tahoe 26 represents a significant leap in forensic granularity, capturing user interactions at the menu level. This data is invaluable for reconstructing attack paths, such as data exfiltration sequences where a user compresses files before deletion. The artifact's reliance on menu text means it may miss context for generic commands, necessitating correlation with file system logs. Mitigation involves incorporating ccl-segb or similar tools into forensic workflows to parse SEGB-encapsulated protobuf data. The artifact's presence in restricted directories suggests Apple's intent for privacy, yet its forensic utility outweighs potential privacy concerns for investigators.

Palo Alto Unit 42 · 2026-06-12 · Case Studies: macOS Tahoe 26's App.MenuItem Biome stream logs user menu selections, enabling forensic reconstruction of intent and workflow.

The Outsider operation exemplifies the growing sophistication of cybercrime-as-a-service models, leveraging AI to automate and scale phishing campaigns. Attackers used phishing kits to create fake sites mimicking trusted brands, bypassing multi-factor authentication (MFA) by requesting SMS, PIN, email, and app verifications. The operation's infrastructure included domains hosted on U.S. providers, payment wallets, and a Shopify storefront, highlighting the blend of legitimate and illicit services. Mitigation requires enhanced email filtering, MFA resilience testing, and user awareness training to recognize phishing lures. Collaboration between tech firms and law enforcement, as seen here, is critical to disrupt such networks.

CyberScoop · 2026-06-12 · Incidents: FBI and partners dismantled a China-based cybercrime network causing $1.9B in losses through AI-powered phishing kits.

The attack path leverages Splunk's PostgreSQL Sidecar Service endpoint (/v1/postgres/recovery/backup), which improperly handles authentication and file operations. By sending a crafted request with a malicious backup file path, attackers can write arbitrary files to the system, including Python scripts in Splunk's webroot. This leads to RCE when the file is accessed. The vulnerability primarily affects on-premise Splunk Enterprise installations where the PostgreSQL Sidecar Service is enabled. Mitigation involves disabling the service if unused, applying patches immediately, and restricting network access to Splunk management interfaces. Organizations should also monitor for unexpected file writes in Splunk directories.

watchTowr Labs · 2026-06-12 · Vulnerability: Splunk Enterprise's PostgreSQL Sidecar Service vulnerable to pre-auth RCE (CVE-2026-20253) with CVSS 9.8, allowing arbitrary command execution via crafted backup requests.

The attack path begins with exploitation of CVE-2026-35273, an unauthenticated RCE flaw in PeopleSoft's EMHub service, which manages agents across environments. ShinyHunters leveraged this to compromise 300+ instances, exfiltrating sensitive data, including student records. The disproportionate impact on universities stems from their reliance on PeopleSoft for administrative functions. Mandiant and GTIG's findings highlight the importance of immediate mitigations: patching, disabling EMHub, and restricting external access. While WAFs offered partial protection, they are not a long-term solution. The incident underscores the need for proactive vulnerability management in critical infrastructure.

Dark Reading · 2026-06-12 · Incidents: ShinyHunters exploited a critical Oracle PeopleSoft zero-day (CVE-2026-35273) to steal data from over 100 organizations, targeting higher education institutions.

The attack vector capitalizes on the AUR's reliance on package history rather than current maintainer trustworthiness. By adopting orphaned packages, attackers inserted malicious build scripts (PKGBUILD/.install) to execute npm install atomic-lockfile, which triggers a preinstall hook running a Linux ELF binary (deps). The payload, a Rust infostealer, targets developer secrets and, with root access, loads an eBPF rootkit to conceal its presence. The rootkit manipulates /sys/fs/bpf/ to hide processes, namespaces, and inodes. Detection requires auditing build histories for atomic-lockfile or js-digest installations and inspecting systemd services and BPF maps. Mitigation demands host reinstallation if rootkit involvement is suspected, as system integrity cannot be assured otherwise. This incident underscores the risks of community-maintained repositories and the need for vigilant package vetting.

The Hacker News · 2026-06-12 · Incidents: Attackers hijacked 400+ Arch Linux AUR packages to deploy an infostealer and eBPF rootkit by exploiting abandoned projects and spoofing commit metadata.

The Outsider network exemplifies the evolving sophistication of phishing operations, leveraging AI to automate and scale attacks. The attack path begins with AI-generated HTML for counterfeit websites, distributed via bulk SMS campaigns. Affected scope includes 9,000 fake websites and 1.59 million URLs, with 2.5 million messages sent to Android users in a two-week period. Mitigation involves multi-layered defenses: telecom providers blocking malicious messages, AI platforms monitoring for abuse, and user education on recognizing phishing attempts. The modular structure of the network—developers, data brokers, spammers, and theft groups—demonstrates the industrialization of cybercrime, requiring coordinated legal and technical responses.

The Hacker News · 2026-06-12 · Incidents: Google sues Chinese cybercrime network for using Gemini AI in smishing attacks, targeting over 100,000 victims via phishing-as-a-service kits.

The takedown of CFAKE.com and SOCFAKE.com reveals the sophisticated infrastructure behind deepfake porn networks, leveraging international domains and cryptocurrency to evade detection. Attackers exploited AI tools to create and distribute nonconsensual content, targeting high-profile individuals. The operation highlights the critical role of cross-border collaboration in combating such crimes. Mitigation strategies include stricter platform monitoring, enhanced legal penalties, and public awareness campaigns to deter creators and users. Organizations should also implement digital watermarking and content authentication tools to identify and flag deepfake material proactively.

CyberScoop · 2026-06-12 · Incidents: International authorities shut down deepfake porn sites CFAKE.com and SOCFAKE.com, arresting a suspect and seizing domains under the TAKE IT DOWN Act.

The case underscores the persistent threat posed by ransomware groups, even after disbandment. Conti's attack path typically involved deploying custom malware to encrypt victim data, followed by extortion demands. The group's resilience was evident as it quickly rebranded into new entities post-disruption. Affected scope included 1,000+ victims across 47 U.S. states and 31 countries, with critical infrastructure and government agencies heavily targeted. Mitigation strategies should include robust endpoint protection, regular backups, and employee training to recognize phishing attempts. Additionally, organizations should monitor for IOCs linked to Conti's successor groups, such as Black Basta and BlackSuit, which continue to employ similar tactics.

CyberScoop · 2026-06-12 · Incidents: Conti ransomware member pleads guilty, faces 20 years for global attacks extorting over $150 million.

The 23andMe breach underscores the vulnerabilities inherent in handling highly sensitive genetic data. Attackers likely exploited weak authentication or API flaws to access DNA Relatives and Family Tree profiles, exposing 5.5 million and 14.1 million users respectively. The breach's impact is compounded by the irreversible nature of genetic data exposure, which can lead to long-term privacy risks. Mitigation strategies should include robust encryption, multi-factor authentication, and regular security audits for genetic testing companies. Additionally, organizations storing sensitive data must prioritize breach response plans to minimize fallout and legal liabilities.

The Record by Recorded Future · 2026-06-12 · Incidents: 23andMe data breach victims to receive $47 million settlement after 7 million customers' genetic and personal data was stolen and leaked.

The lapse of Section 702 creates immediate operational gaps for U.S. intelligence agencies, particularly in monitoring foreign adversaries and terrorist activities. The program's expiration removes warrantless access to digital communications of overseas targets, a key tool for real-time threat detection. Mitigation efforts should focus on interim measures, such as leveraging alternative intelligence-sharing frameworks with allies and accelerating legislative reforms to restore oversight and authority. The political impasse highlights the need for bipartisan consensus on balancing national security with privacy concerns, ensuring future renewals include safeguards against misuse.

The Record by Recorded Future · 2026-06-12 · Policy: Section 702 of FISA has lapsed due to legislative deadlock, disrupting U.S. intelligence gathering on foreign threats.

The allegations against IBM and AT&T highlight systemic issues in corporate transparency and compliance with breach disclosure laws, potentially undermining trust in critical infrastructure providers. The University of Oxford breach demonstrates the persistent vulnerability of educational institutions to targeted attacks, particularly through third-party services. Google's layoffs in its cybersecurity division raise concerns about the prioritization of security investments within major tech firms. Microsoft's AI playbook is a proactive step toward addressing the unique challenges posed by AI in security incidents. Nokia's Deepfield Genome Shield represents an innovative approach to mitigating large-scale DDoS attacks, which are increasingly leveraging residential proxy botnets. ENISA's focus on collective resilience reflects the growing recognition of transnational cyber threats and the need for coordinated responses. The takedown of AudiA6 underscores the importance of international law enforcement collaboration in disrupting cybercrime ecosystems.

SecurityWeek · 2026-06-12 · Incidents: Key developments include IBM/AT&T hack cover-up allegations, Oxford data breach, Google layoffs, Microsoft AI playbook, Nokia DDoS defense, ENISA resilience focus, and AudiA6 takedown.

The attack leverages CVE-2026-35273, a critical RCE flaw in Oracle PeopleSoft PeopleTools, to gain unauthorized access to servers. Attackers bypass authentication, execute arbitrary code, and exfiltrate sensitive data. The education sector, representing 68% of victims, is disproportionately affected due to widespread PeopleSoft usage. Oracle's delayed response and lack of a patch exacerbate the risk. Mitigation involves isolating affected systems, applying interim workarounds, and monitoring for suspicious activity. Organizations should prioritize patching similar vulnerabilities in Oracle products, given ShinyHunters' history of targeting unpatched systems. The campaign's persistence underscores the need for proactive threat hunting and collaboration with threat intelligence providers.

CyberScoop · 2026-06-12 · Incidents: ShinyHunters exploits Oracle PeopleSoft zero-day (CVE-2026-35273) to extort universities, with over 100 organizations potentially compromised.

The breach underscores systemic failures in Coupang's security posture, particularly in key management and access controls. A former employee exploited insider knowledge to steal a signing key, enabling unauthorized access to customer data. The attack path involved a test run in January 2025, followed by systematic exploitation from April. The scope extended beyond registered members to include non-members whose data was stored without consent. Mitigation should focus on robust key management, regular access reviews, and ensuring the independence of privacy officers. The incident highlights the importance of transparency in breach notifications and the legal risks of mishandling personal data.

The Record by Recorded Future · 2026-06-12 · Incidents: South Korea fines Coupang $409 million for a data breach affecting 33.7 million customers due to basic security failures and mishandling of the incident.

The article underscores the inadequacy of traditional perimeter defenses in cloud-native environments, where attackers can exploit east-west traffic for lateral movement once inside the cluster. Istio Ambient Mesh is presented as a solution to enforce zero-trust principles at the service layer, reducing reliance on perimeter controls. The human element remains critical, as open-source communities and security teams must interpret conflicting records and resolve automation failures. Supply chain vulnerabilities and standing privileges are highlighted as high-risk areas, with lessons from Shai Halud and TeamPCP campaigns providing actionable insights. Mitigations include adopting zero-trust frameworks, enhancing supply chain security, and ensuring human oversight in automated systems.

GitGuardian Blog · 2026-06-12 · Events: Trust in cloud-native systems requires continuous validation, human oversight, and zero-trust architectures to mitigate lateral movement and supply chain risks.

The release of Claude Fable 5 and Mythos 5 underscores the dual-use nature of advanced AI models. Mythos 5, with its enhanced exploit discovery capabilities, poses a significant threat if misused, though Anthropic's Project Glasswing aims to mitigate this through restricted access. However, the assumption that adversarial groups already possess similar capabilities is reasonable, given historical precedents. Attack paths could involve insider threats or compromised credentials within partner organizations. Defenders should focus on reducing attack surfaces by implementing strict egress filtering, continuous vulnerability assessments, and integrating AI-driven security tools to keep pace with potential AI-augmented attacks. The lack of defensive research access to Fable 5 further complicates the landscape, as it limits the development of countermeasures.

Dark Reading · 2026-06-12 · Vulnerability: Claude Fable 5 and Mythos 5 models maintain existing security risks, urging defenders to prioritize foundational security measures.

The article highlights a critical operational challenge in cybersecurity: maintaining scalable, real-time scanning infrastructure. Cloudflare's initial system suffered from bottlenecks in Kafka message processing and API timeouts, which are common pain points in distributed systems. Their solution involved deep system analysis rather than brute-force resource allocation, focusing on optimizing existing architecture. This approach not only resolved immediate scalability issues but also enabled new features like granular on-demand scans. The case underscores the importance of system observability (metrics, logs) and iterative optimization in security tooling. For enterprises, the key takeaway is that scaling security operations requires both architectural foresight and continuous performance tuning.

Cloudflare Blog · 2026-06-12 · Case Studies: Cloudflare scaled security scanning throughput 10x to enable automatic scans for all accounts and reduce risk detection windows.

The release of Claude Fable 5 highlights the escalating arms race in AI-driven cybersecurity. Attack paths now include AI-assisted reconnaissance, exploit development, and lateral movement at machine speed, outpacing human defenders. Affected scope spans enterprises relying on legacy defenses, particularly those without AI-integrated security postures. Mitigations include real-world attack surface testing, AI-enhanced vulnerability patching, and secure coding practices. Proactive governance and tiered access controls are critical to balancing innovation and risk. The defender lifecycle must accelerate to match AI's offensive potential, with a focus on secure-by-design principles.

SecurityWeek · 2026-06-12 · Vulnerability: Claude Fable 5's safeguards and dual-use capabilities underscore the urgent need for defender adaptation against AI-driven hyperattacks.

The Agentjacking attack capitalizes on the implicit trust between AI coding agents and external services like Sentry. By injecting crafted markdown into error events, attackers deceive agents into executing malicious commands. The attack path involves: 1) identifying a target's Sentry DSN, 2) sending a malicious POST request to Sentry's ingest endpoint, and 3) tricking the AI agent into processing the payload as a legitimate resolution step. This method affects organizations using AI coding assistants like Claude Code and Cursor, with Tenet identifying 2,388 exposed organizations. Mitigations include restricting DSN access, implementing strict input validation for AI agents, and monitoring agent behavior for anomalous commands. Sentry's global content filter offers partial protection but fails to address the root architectural flaw.

The Hacker News · 2026-06-12 · Vulnerability: Agentjacking attacks trick AI coding agents into executing malicious code via manipulated Sentry error reports, bypassing traditional security controls.

The CyberCorps program is a linchpin in developing a skilled federal cybersecurity workforce, yet its funding instability under the Trump administration jeopardizes its mission. AI's rapid adoption by threat actors necessitates advanced training, but budget cuts could cripple the program's ability to deliver. Attack paths include AI-driven vulnerability discovery and automated attacks, which CyberCorps graduates are uniquely positioned to counter. Mitigation requires consistent funding to expand scholarships and integrate AI training. The Pentagon's need for 25,000 cyber experts highlights the urgency. Congress's proposed $60-70 million funding is a step forward, but long-term stability is critical.

CyberScoop · 2026-06-12 · Policy: CyberCorps program adapts to AI threats but faces funding cuts, risking U.S. cyber defense readiness.

Handala’s intrusion into Cal Water highlights a calculated attack path: initial compromise of the RTKBase GNSS platform, likely due to weak credentials or unpatched vulnerabilities, followed by lateral movement to the billing system. The exposed data—names, addresses, payment histories, and administrative credentials—poses significant privacy and operational risks. Handala’s toolkit, including custom wipers like win.handala and MBR-overwriting capabilities, suggests a dual intent: data theft and potential disruption. The group’s operational pattern, as seen in the Stryker incident, often escalates from exfiltration to destruction. Defenders must prioritize isolating the RTKBase instance, auditing lateral movement paths, and monitoring for wiper deployment. The lack of public acknowledgment from Cal Water underscores the need for transparent incident response protocols.

SecurityWeek · 2026-06-12 · Incidents: Handala breached Cal Water, exfiltrating PII and credentials via RTKBase, posing a potential precursor to destructive attacks.

The discovery of 'short-sleeve' RSA keys highlights a systemic issue in cryptographic key generation, where biased bits create exploitable patterns. The polynomial-based factorization technique leverages these patterns to break keys efficiently, but only when the bias is significant. The CompleteFTP bug, responsible for many of these keys, was a type mismatch in big-integer code, affecting both RSA and DSA keys. Defenders should prioritize checking and regenerating keys generated by vulnerable versions of CompleteFTP, especially those used in critical infrastructure. The research underscores the importance of rigorous randomness in cryptographic implementations to prevent such vulnerabilities.

Trail of Bits Blog · 2026-06-12 · Vulnerability: Biased RSA key generation in CompleteFTP led to hundreds of factorable 'short-sleeve' keys.

The MDR model, designed for a slower, human-paced threat landscape, is now inadequate against AI-enhanced attacks. Attackers leverage AI to automate phishing, reconnaissance, and malware creation, targeting low-severity alerts that MDR services often deprioritize. Analysis of 25 million alerts reveals that nearly 1% of real threats originate in these overlooked alerts, translating to significant breaches. Organizations must augment their MDR with AI-driven SOCs to ensure full alert coverage. This transition involves integrating AI tools alongside existing MDR contracts, observing their effectiveness, and making informed decisions at renewal. AI SOCs provide forensic-level investigation of every alert, closing the coverage gap that MDRs structurally cannot address.

The Hacker News · 2026-06-12 · Incidents: AI-driven threats expose critical gaps in MDR services, necessitating a shift to AI-driven SOCs for comprehensive alert coverage.

The attack path begins with an SQL injection (CVE-2025-67644) in LangGraph's SQLite checkpoint implementation, allowing attackers to inject malicious metadata filters. This returns a fake checkpoint row containing attacker-controlled serialized data. The subsequent deserialization (CVE-2026-28277) of this data via msgpack triggers arbitrary code execution. The chain requires write access to the checkpoint store and exposure of the get_state_history() endpoint. Self-hosted deployments using SQLite or Redis checkpoints are at risk, while LangSmith-managed platforms are not. Defenders should prioritize patching, segment networks, and treat AI agents as privileged identities due to their elevated access. The findings underscore how traditional vulnerabilities like SQLi gain potency in AI frameworks, where agents often operate with broad permissions.

The Hacker News · 2026-06-12 · Vulnerability: LangGraph's SQLi and unsafe deserialization flaws enable RCE in self-hosted AI agents via checkpoint manipulation.

The CVE-2026-10520 vulnerability in Ivanti Sentry is a severe OS command injection flaw exploitable remotely without authentication, granting root privileges. Attackers target the management port (8443), but exploitation requires external exposure, which is atypical in properly configured environments. Ivanti emphasizes that EPMM-managed appliances use mTLS, rendering the API inaccessible to external actors. Unmanaged instances, though vulnerable, are not production-viable. Organizations should prioritize patching, restrict internet access to the management port, and ensure mTLS is enforced. This vulnerability underscores the importance of network segmentation and strict access controls for management interfaces.

SecurityWeek · 2026-06-12 · Vulnerability: Ivanti Sentry's critical CVE-2026-10520 shows exploitation attempts on honeypots, with patches available and risk reduced by proper configuration.

The Chrome 149 update highlights the persistent threat of memory safety vulnerabilities, particularly use-after-free flaws, which dominate the patched issues. These vulnerabilities can be chained with OS or browser privilege flaws to achieve sandbox escape, a critical escalation vector. Google's mitigation strategies, like MiraclePtr and Rust adoption, reflect long-term defense against such exploits. The spike in reported vulnerabilities, potentially AI-aided, underscores the need for rapid patch deployment. Defenders should prioritize updating Chrome instances immediately, especially in environments where browser-based attacks are prevalent. Monitoring for unusual browser behavior or unexpected process spawns can help detect exploitation attempts.

SecurityWeek · 2026-06-12 · Vulnerability: Chrome 149 patches 28 vulnerabilities, including critical use-after-free flaws, amid a surge in exploits and Google's shift to Rust for memory safety.

Sniper Dz operated as a sophisticated PhaaS platform, offering phishing kits, hosting, and operational support, primarily targeting tech, social media, and streaming platforms. The attack path involved impersonating popular brands and government entities through convincing imitation websites, harvesting credentials and personal data. The platform's free infrastructure model democratized phishing, enabling low-skilled threat actors to launch campaigns. Mitigation includes enhanced user education on phishing indicators, multi-factor authentication (MFA) enforcement, and continuous monitoring for domain impersonation. Organizations should also collaborate with law enforcement to disrupt similar platforms and share threat intelligence.

The Hacker News · 2026-06-12 · Incidents: INTERPOL disrupts Sniper Dz, a decade-old PhaaS platform, arresting its administrator and seizing infrastructure used in global phishing campaigns.

The alleged jailbreak relies on multi-agent prompting to exploit conversational refusals, a known LLM limitation, rather than bypassing core safety classifiers. Attack paths involve sophisticated prompt engineering to elicit responses on restricted topics, but Anthropic's layered defenses—including fallback to Claude Opus 4.8 for high-risk queries—remain intact. Scope is limited to general information retrieval, not actionable exploit or weaponization guidance. Mitigations include continuous red-teaming, classifier updates, and monitoring for adversarial prompting patterns. Defenders should treat this as a proof-of-concept for LLM probing, not a systemic compromise.

SecurityWeek · 2026-06-12 · Vulnerability: Anthropic refutes claims of a Claude Fable 5 jailbreak, citing robust safeguards and limited impact of alleged bypass.

The AudiA6 operation underscores the sophistication of modern crypto laundering services, which leverage chain-hopping, decentralized exchanges, and mixer-as-a-service platforms to obscure illicit funds. Attack paths often involve recruiting Russian-speaking intermediaries and using controlled domains to register mule accounts. Defenders should monitor for transactions linked to the listed domains and enhance KYC protocols. The seizure of 25 domains and 30 servers demonstrates the scale of infrastructure supporting such services. Mitigations include stricter exchange onboarding, blockchain analytics to trace illicit flows, and international collaboration to disrupt criminal networks.

The Hacker News · 2026-06-12 · Incidents: Europol dismantles AudiA6, a crypto laundering service used by ransomware gangs, seizing assets and arresting key operators.

The vulnerability stems from a fundamental design flaw in Check Point's IKEv1 implementation, where the gateway blindly accepts client-specified certificate validation levels. Attackers can force 'no verification' mode, granting full VPN access without valid credentials. The attack path is versatile, working across UDP/500 (traditional IKE) and TCP/443 (Visitor Mode via TCPT protocol), bypassing common perimeter filtering. All certificate authentication modes (standard, enrollment, mixed) are vulnerable; only XAUTH password-based authentication remains secure. Mitigation requires immediate application of Check Point's June 2026 hotfixes, disabling IKEv1 where possible, and monitoring for anomalous VPN connections using watchTowr's Detection Artefact Generator. Network defenders should prioritize this patch given the CISA KEV listing and confirmed ransomware linkages.

watchTowr Labs · 2026-06-12 · Vulnerability: Check Point VPNs vulnerable to IKEv1 authentication bypass (CVE-2026-50751), actively exploited by ransomware groups.