Collection of cybersecurity reports, surveys, research papers, market studies, and industry intelligence.
726 articles
Acronis platform data shows that script reuse and AI generation are distinct scaling approaches. Reused templates can spread defects across tenants, while AI-generated scripts require review and verification before running with endpoint privileges.
VulnCheck found 14 of 1,061 AI-assisted vulnerabilities disclosed in the first half of 2026 were exploited, a rate in line with other vulnerabilities. Exploitation after CVE publication accelerated from 120 days to 80.
CertiK identified 52 wrench attacks worldwide in the first half of 2026, up 33% year over year. Reported losses reached $124 million, though CertiK cautioned that figure reflects financial scale, not necessarily realized criminal profits.
Netskope’s AI report, covering June 2025–July 2026, identifies rising downstream data-access risks from MCP, agentic coding tools as potential execution vectors, and weaponized AI supply-chain threats.
CloudSEK reports rising France-related dark-web activity, driven largely by exposed credentials and infostealer logs, alongside ransomware claims targeting local governments and geopolitically motivated hacktivism.
Netskope says it detected and blocked over 28,000 World Cup-themed threats across 1,000+ organizations in 2026. Users reaching malicious content peaked at nearly six times the pre-tournament average; lures included job scams, fake streams and infostealers.
Eclypsium’s inaugural July 2026 InfraTrust Pulse reviewed 61 infrastructure advisories from 14 vendors, prioritizing exposure and exploitability over CVSS. It highlighted actively exploited SonicWall SMA1000 flaws and Fortinet FortiSandbox issues added to CISA’s KEV.
JPCERT/CC’s TSUBAME sensors observed changing IoT scanning activity: Mirai-like 23/TCP traffic from Japan declined during FY2025, while scans from other compromised devices increased. No unusual activity requiring a special alert was seen in Q1 2026.
The UK AI Security Institute found that all tested OpenAI and Anthropic models attempted to cheat in cyber evaluations. One model ran code on an external service to try to access evaluation infrastructure; AISI reported no data leaks or damage.
SentinelLABS assesses that Iran-linked cyber activity remains focused on persistent access, espionage, persona-led coercion, and exposed OT systems, while public claims often exceed verified effects and visibility inside Iran remains limited.
Flare analyzed 2,889 underground posts and found carders increasingly assess residential proxies by reputation and precise location, while pairing them with browser and identity signals. The article outlines fraud-detection clues for defenders.
CYFIRMA examines six ways AI can enable industrial espionage, from agentic intrusions and synthetic identities to model theft and voice cloning, while distinguishing reported cases from speculative risks.
Lava Labs’ report identifies 10 security risks in AI data centers, arguing that rapid construction and multi-tenant GPU workloads expand the exploitability and blast radius of infrastructure weaknesses. It ranks risks by severity, likelihood, impact and detectability.
Aikido reports that GPT-5.6 and Kimi K3 each rediscovered 23 of 26 known CVEs in its code-analysis benchmark. Repeated runs of cheaper models often found more bugs than a single run of a pricier model.
Recorded Future finds that AI is accelerating Iran’s existing cyber and influence operations, while evidence of AI’s role in military operations remains limited or unconfirmed. It recommends resilience against combined cyber, information, and physical threats.
A June 2026 analysis of financial-sector threats found phishing prominent in initial attacks, malware in later stages, and account theft, data sales, ransomware claims, and access listings across dark-web sources.
AhnLab’s June 2026 report describes infostealer distribution through file-sharing sites, search poisoning and email, along with macOS campaigns using ClickFix, Bash scripts and a smart-contract-based method for retrieving C2 addresses.
Check Point Research’s 2026 report says AI is increasingly used to operate intrusions, build malware, enable social engineering, and expose enterprise data, while indirect prompt-injection detections rose roughly fivefold from March to May.
Acronis’s June 2026 threat report, based on data from more than one million endpoints, tracks malware and URL detection trends and highlights claimed ShinyHunters data theft involving a PeopleSoft zero-day.
ENISA’s survey of 194 organisations across 31 countries found gaps between SMEs’ awareness of the Cyber Resilience Act and practical readiness, with microcompanies generally less mature and seeking templates and financial support.
CloudSEK identified fake Tomorrowland ticket and travel sites that can steal money, personal information, and payment details. The report details their impersonation tactics, checkout flows, infrastructure, and advice for festival-goers.
Acronis reports that session recording adoption grew broadly in the two months after its April 2026 launch, with weekly recording volume up about 178% from the first full week to its peak and several markets reaching double-digit adoption.
A threat-intelligence report describes a decentralized pro-Iran cyber ecosystem that relies mainly on DDoS, defacement, hack-and-leak claims and propaganda to create disruption and psychological pressure, and outlines defensive priorities.
Recorded Future identified 59 high-impact vulnerabilities in June 2026; 23 were in CISA’s KEV catalog, and 56 were reported as actively exploited. The report also highlights malware-linked campaigns, widespread RCE exposure, and public PoCs.
