Issue 15
Critical Cisco SD-WAN flaw exploited in zero-day attacks since 2023
A critical Cisco SD-WAN vulnerability has reportedly been exploited in zero-day attacks since 2023. Eleven sources reported the news in the middle of the week, on 25 February.
Who is affected and how
- security leader
- Organisations that use SD-WAN need to check whether their infrastructure is affected and search back for traces of compromise going as far as 2023.
- security operations
- Patching and reviewing network device logs is urgent, because the exploitation has been going on for a long time.
University of Mississippi Medical Center reopens clinics after ransomware attack
The University of Mississippi Medical Center's clinics across the state closed because of a ransomware attack, then reopened on 23 February. The digests give no details of the circumstances of the…
security leaderThe incident is a test of healthcare institutions' business continuity plans: the loss of clinics directly affects patient care.citizenPatient care was temporarily interrupted by the attack.vulnerability · United StatesCISA adds two Roundcube vulnerabilities to its catalog of actively exploited flaws
CISA has expanded its Known Exploited Vulnerabilities (KEV) catalog with flaws affecting the Roundcube webmail platform. The digest points to 20 February, and ThaiCERT also confirmed the entry.
security leaderA KEV entry means a remediation deadline and a priority; webmail exposure needs to be assessed.security operationsServers running Roundcube need to be patched urgently, because CISA indicates active exploitation.law enforcement · United StatesUS Treasury sanctions Russian firm that allegedly stole and sold US cyber tools
The US Department of the Treasury has imposed sanctions on a Russian firm that it says stole and sold US cyber tools. The move coincides with an FBI investigation into a former L3Harris employee who…
security leaderInsider-threat risk at defence subcontractors and the leakage of exploit tools are the subject of a specific sanctions case.public-sector decision-makerThe sanction shows the state's toolkit for acting against players in the exploit market.incident · globalAnthropic says DeepSeek, Moonshot AI and MiniMax used 16 million Claude queries to copy its model
Anthropic said on Monday that it had identified "industrial-scale campaigns" run by DeepSeek, Moonshot AI and MiniMax. According to Anthropic, the companies used 16 million Claude queries to copy the…
security leaderModel extraction (distillation) is a risk to the intellectual property of AI models.developerDevelopers who offer AI services through an API need to prepare to detect and limit anomalous, high-volume queries.journalist and researcherThe claim comes from Anthropic; the three companies' response is not included in the summary.Hungary
More key stories of the week
Google disrupts China-linked campaign targeting telecoms and government organisations
Google says it disrupted a major China-linked cyber-espionage campaign that targeted telecoms and government organisations. The campaign reportedly affected 53 organisations worldwide.
security leaderThe telecoms and government sectors face a heightened espionage risk.public-sector decision-makerThe targeting of government bodies should be treated as a state-level risk.incidentHackers reportedly used Anthropic's Claude to steal 150 GB of Mexican government data
Attackers reportedly used Anthropic's Claude AI tool to steal 150 GB of Mexican government data. The story appeared in five sources on 26 February.
security leaderThe case suggests that commercial AI tools can also speed up attackers' work.public-sector decision-makerProtecting government data now has to account for AI-assisted intrusion.incident · globalAmazon says Russian-speaking hackers used generative AI to compromise 600 firewalls
Russian-speaking hackers used generative AI tools to compromise 600 firewalls, according to Amazon. The report appeared on 23 February.
security leaderAI can also increase the scale of attacks by less skilled attackers.security operationsAccess controls and configuration of perimeter firewalls need to be rechecked.bűnüldözés · Európai UnióEuropol-led Project Compass leads to 30 arrests of The Com members
Europol's year-long Project Compass operation led to 30 arrests and linked 179 suspects to The Com, an online criminal community. Authorities from 28 countries took part.
security leaderAction against The Com's extortion and ransomware networks shows international cooperation against decentralised criminal groups.journalist and researcherThe operation's figures come from official statements.attribution · globalReports link North Korea's Lazarus group to Medusa ransomware attacks
Six sources reported during the week that North Korea's Lazarus group has been linked to Medusa ransomware attacks. The summary does not name who made the finding.
security leaderThe line between state-backed and financially motivated attacks may blur when assessing ransomware risk.journalist and researcherThe link is a claim made in the articles; the summary gives no detailed evidence.regulation · European UnionNIS2 and Ireland's national cyber security bill: what management boards must know
An IAPP analysis published on 25 February discusses the responsibilities of management bodies under Ireland's National Cyber Security Bill and the NIS2 directive. The available excerpt gives no…
security leaderBuilding a compliance programme requires the involvement of senior management.business leaderNIS2 stresses the liability of management bodies.incident · United StatesConduent ransomware breach: estimated number of affected people rises from 10 million to 25 million
A ransomware attack on Conduent put the data of more than 15 million Americans at risk, and later reporting puts the number above 25 million. The articles were published between 24 and 26 February.
business leaderThe growth in the estimated number of victims shows how difficult incident communication is.citizenAffected people's personal data may have been exposed, so they should watch for suspicious messages and fraud attempts.research · globalVoidLink analysis: a Linux malware framework built with AI
In January 2026, researchers identified a Linux malware framework called VoidLink. The analysis says its significance lies in how it was built, not only in the fact that it targets Linux.
security leaderProtection of Linux servers and cloud infrastructure must take AI-developed malware into account.security operationsDetection coverage on Linux endpoints should be reviewed.case · United StatesMarquis sues SonicWall over backup breach that led to ransomware attack
Marquis has filed a lawsuit against SonicWall over a breach involving backups that led to a ransomware attack. The news appeared on 25 February.
security leaderThe supplier risk of cloud backup services should be managed through both contracts and technical controls.business leaderThe case is an example of enforcing supplier liability through the courts.attribution · globalNorth Korean hackers run fake IT worker and "Contagious Interview" campaigns
State-backed North Korean hackers are running large-scale fake IT worker and "Contagious Interview" campaigns that exploit developer hiring processes. The reports appeared on 23 February.
developerCode and tasks received during job interviews should only be run in an isolated environment.business leaderVerifying identity when hiring remote workers is a security issue.attribúció · Európai UnióAPT28 linked to attacks on European organisations using webhook-based macro malware
According to HackenPost, the APT28 group targeted European organisations with webhook-based macro malware. The story appeared on 23 February.
security leaderThe campaign targets European organisations, so it should be included in sector threat monitoring.security operationsRestricting macro execution and monitoring outbound traffic to webhook services can reduce the risk.report · globalCrowdStrike warns AI-powered cyber-attacks rose significantly over the past year
CrowdStrike warned that AI-powered cyber-attacks have increased significantly over the past year. The news appeared on 24 February.
security leaderThe vendor data also justifies updating the threat model for AI-driven attacks.business leaderAI-based attacks are emerging as a question of both budget and preparedness.In brief
Weekly email
One email a week with the main stories from the Agent Age editions. You confirm by email and can unsubscribe in one click.
Almost done: we sent you an email. Open it and press the confirm link. If it does not arrive within a few minutes, check your spam folder.
Please enter a valid email address and tick the consent box.
Something went wrong. Please try again later.