Issue 53 · Nov 16–22, 2026Agent Age News magazine · Other editions · Search · Method · HU / ENG

Incidents and influence attempts in the age of agents

Issue 29

Jun 1–7. 2026 · 43 stories · 5 of them Hungarian and regionalPage last updated: Nov 21, 2026
Viewpoint

incident · global

Red Hat's official npm packages compromised to spread credential-stealing worm

Red Hat's official npm accounts were compromised and used to push a malicious, self-propagating worm; the reports appeared on 1 June. One account says the attack affected more than 30 packages, and the worm steals sensitive data.

2026. Jun 1. · 12 sources

Who is affected and how

security leader
The malware arrived through an official publisher channel, so supply-chain controls and package verification need to be reassessed.
security operations
The worm spreads from machine to machine, so the affected packages must be located across environments and compromised machines isolated.
developer
Anyone who installed the affected Red Hat npm packages may have had credentials exposed on their machine and build environment, so they should rotate secrets and review their dependencies.
incident · global

Hackers tricked Meta AI support chatbot into handing over celebrity Instagram accounts

Attackers got Meta's AI support chatbot to help them take over celebrity Instagram accounts; the Guardian cites the hijacking of the Obama White House account (1–3 June). Analyses point to the…

security leaderAn AI chatbot in an account-recovery or customer-support process can become an attack surface, so its permissions and identity-verification steps should be reviewed.developerA support system that operates as an agent or chatbot must be designed so that it does not carry out account actions when deceived.citizenSocial media account recovery and support channels can also be targets, so two-factor protection and caution are warranted.
vulnerability · global

CISA adds critical Palo Alto Networks firewall flaw to KEV catalog

On 1 June, CISA added a critical vulnerability in Palo Alto Networks firewalls to its Known Exploited Vulnerabilities catalog. The company and researchers are also warning of exploitation.

security leaderBecause the firewall flaw is being actively exploited, the exposure of perimeter devices and the patching process should be reviewed out of cycle.security operationsPerimeter firewalls must be patched or mitigated urgently, and checked for signs of compromise.public-sector decision-makerThe KEV listing signals that patching cannot be postponed at organisations using the affected devices.
vulnerability · global

Critical Windows Netlogon RCE flaw now exploited in attacks

Reports say attackers are already exploiting a critical remote code execution flaw in Windows Netlogon (June 1). The vulnerability affects domain infrastructure.

security leaderAn already exploited RCE flaw in domain infrastructure calls for incident readiness and patch escalation.security operationsDomain controllers must be patched out of cycle, and Netlogon traffic should be monitored.
vulnerability · global

Cisco warns of unpatched SD-WAN zero-day exploited in attacks

On June 5, Cisco warned that an unpatched SD-WAN zero-day vulnerability is being exploited in attacks. According to reports, no fix is yet available for the flaw.

security leaderAn unpatched flaw exploited on a network backbone device calls for a risk management decision and heightened monitoring.security operationsWithout a patch, the mitigation recommended by Cisco should be applied and SD-WAN device logs monitored.

Hungary

regulationAbout 1,000 Hungarian companies must complete NIS2 cybersecurity audit by end of Junesecurity leaderPreparation, documentation and the remedying of gaps for the audit must be completed by June 30.business leaderThe company concerned must finish the audit by the deadline, or it faces a compliance risk.public-sector decision-makerCompliance by the obligated businesses is the yardstick for how NIS2 rules are being applied in Hungary.influence operationPéter Magyar calls on Viktor Orbán to stop Fidesz disinformation campaignjournalist and researcherThe disinformation allegation cannot be reported as fact without verifying the campaign's specific content and measured reach.citizenPoliticians' mutual disinformation accusations are best assessed with the source identified.incidentFinland: four suspects in sabotage of undersea cables to Estoniasecurity leaderNetwork operators should review physical route redundancy and readiness for cable outages.public-sector decision-makerProtecting undersea cables is a matter of national and EU infrastructure security.caseCompanies close to Hungary's governing system can gain insight into critical firms' IT vulnerabilitiesbusiness leaderBusinesses performing critical tasks need to clarify who gets access to vulnerability data.public-sector decision-makerClear rules are needed on who may access critical infrastructure vulnerability data.journalist and researcherThe scope of access and its justification require further fact-finding.reportFraud is less frequent in Debrecen, but vigilance must not lapsecitizenEven as local fraud figures improve, caution toward unknown approaches remains warranted.

More key stories of the week

vulnerability · United States

CISA adds actively exploited Mirasvit Magento extension flaw to KEV catalog after remote code execution attacks

According to 4 June reports, CISA added a flaw in Mirasvit's Magento extension to its Known Exploited Vulnerabilities catalog after active remote code execution attacks. Listing means affected…

security operationsMagento-based online shops should immediately check for the Mirasvit extension and either patch or disable it.business leaderBecause of the risk of shop downtime and customer data exposure, patching should be treated as a business priority.
tool · global

OpenAI introduces Lockdown Mode to defend against prompt injection attacks

OpenAI introduced Lockdown Mode on 6 June, a feature designed to counter prompt injection attacks. The move is a vendor response to one of the main types of attack on AI systems.

security leaderWhen enterprises adopt AI tools, built-in protection against prompt injection can be a procurement and risk consideration.developerFor applications built on agents and AI integrations, the hardened mode can serve as a model for restricting permissions.
vulnerability · global

Google fixes one actively exploited Android zero-day and 124 other flaws

Google's Android update, released on June 2, fixes one actively exploited zero-day vulnerability and 124 flaws in total. The fix makes prompt updating of devices advisable.

security operationsThe update must be enforced on corporate Android fleets, because one of the flaws is already being exploited.citizenAndroid devices should be updated as soon as possible.
incident · United States

Amtrak data breach reportedly affects millions of customer records

Reports of 1 June say a data exposure linked to Amtrak may affect millions of customer records. It poses a privacy risk to those affected.

business leaderFailures in customer data handling create legal and trust risks.citizenAffected passengers should watch for phishing and fraud that misuse their personal data.
incident

Carnival data breach affects nearly 6 million travelers

Reports of 1 June say a Carnival data breach exposed the data of nearly 6 million cruise travelers. The incident involves a large volume of customer data.

business leaderFor companies that manage very large customer bases, a data breach carries legal and reputational risk.citizenAffected travelers should expect targeted phishing and fraud.
case · European Union

Anthropic offers EU access to its Mythos hacking technology

Anthropic invited the European Union on June 1 to access Mythos, a technology with hacking capabilities. The move bears on the offensive and defensive use of AI.

security leaderWider availability of AI-driven vulnerability discovery may change both defensive and threat models.public-sector decision-makerEU institutions must decide how to obtain access to an AI-based offensive capability and on what terms.
incident

DentaQuest data breach exposes accounts of 2.6 million people

A data breach at DentaQuest affected 2.6 million accounts, according to reports published on 4 June. It may involve sensitive data from the health insurance sector.

business leaderFor companies handling health data, a breach brings notification and liability consequences.citizenThose affected should watch for misuse of their personal and health data.
case · United States

How NIST mismanaged the National Vulnerability Database

An analysis published on 1 June examines how NIST mismanaged the National Vulnerability Database (NVD). The details are not known from the available headline.

security operationsFor vulnerability management that relies on the NVD, delays and gaps can pose a risk.journalist and researcherOperational problems at the central source of vulnerability data affect the reliability of research and disclosure.
law enforcement · United States

Meta anti-scam operation removes 1.4 million accounts, leads to 63 arrests

In a joint operation with Microsoft, SpaceX and the US Department of Justice, Meta removed more than a million scam accounts (3 June). According to Fox News, the operation with the FBI led to 63…

journalist and researcherThe figures should be traced back to statements by Meta and the authorities.citizenAction against scam networks may reduce exposure, but caution on social platforms is still needed.
report · United States

CISA and partners urge hardening of automatic tank gauge systems

On 2 June, CISA and its partners called for the hardening of automatic tank gauge (ATG) systems. The guidance concerns industrial systems used in fuel storage.

security leaderIndustrial control systems for fuel storage are part of cyber-physical risk management.security operationsATG systems should be isolated from the internet and hardened in line with CISA's guidance.
vulnerability · global

VS Code zero-day lets attackers steal GitHub tokens with one click

Reports from June 3 say a zero-day flaw in VS Code allows attackers to steal GitHub tokens with a single click. The flaw affects developer environments.

security leaderHandling of developer-tool tokens is a supply chain risk, and without a patch mitigation is needed.developerStolen GitHub tokens can give access to repositories and CI, so token permissions and validity should be checked.
vulnerability · United States

CISA: attackers exploit SolarWinds Serv-U flaw to crash servers

According to a CISA warning of June 5, attackers are actively exploiting a SolarWinds Serv-U vulnerability that can crash servers. The flaw affects file transfer servers.

security operationsServ-U instances should be patched, because the flaw can cause service outages.

In brief

attributionChina-linked TA4922 extends phishing attacks to Europesecurity leaderThe campaign tests how well organisations in the named countries are prepared against phishing.journalist and researcherThe attribution is an editorial claim, so the source for any link between the group and the Chinese state should be checked.reportENISA NIS360 report: cybersecurity maturity rises in critical sectors, but unevenlysecurity leaderSectoral maturity indicators provide a benchmark for judging one's own cybersecurity level.public-sector decision-makerThe report can serve as a basis for more targeted sectoral and NIS2 policies.vulnerabilityCISA adds two exploited Microsoft Defender flaws to its KEV catalogsecurity leaderExploitation of endpoint protection flaws calls the reliability of the defensive layer into question, so patch progress must be tracked.security operationsPatching of the two CVEs should be verified on endpoints, because the security software itself is the target.researchSophos uncovers AI-powered malware lab built to evade EDRsecurity leaderEDR-based defenses should be reinforced with additional detection layers and threat hunting.security operationsMalware generated or refined with AI may test both signature-based and behavior-based detection.researchPro-Russian disinformation pollutes AI chatbot responsesjournalist and researcherMeasuring the sources and contamination of chatbot responses requires methodological checking.citizenAI chatbot answers on political or geopolitical topics cannot be considered reliable on their own.regulationTrump's AI executive order aims to strengthen federal cyber defensessecurity leaderThe order may also shape expectations for federal suppliers and AI-based defense.public-sector decision-makerThe order sets a federal cyber defense direction suited to the AI era, which could serve as a reference for other governments.reportAI is making cyberattacks more autonomous and harder to assesssecurity leaderMapping AI-enabled attacker techniques in an ATT&CK-like way can serve as a threat-modeling tool.journalist and researcherThe degree of autonomy and how assessable the attacks are should be examined with data, independent of the claims made.vulnerabilityChrome 149 update patches 429 vulnerabilitiessecurity operationsThe update to Chrome 149 should be scheduled and verified on managed machines.citizenThe browser should be updated if this does not happen automatically.attributionVeritasium: Russian satellite may have jammed GPS over Europesecurity operationsGPS outages affect the operation of systems that depend on navigation and time synchronisation.journalist and researcherThe attribution of satellite jamming needs to be backed by further sources.incidentKelp DAO hacker launders about $220 million in stolen fundssecurity leaderLarge thefts from crypto platforms are hard to trace because the stolen value moves quickly.journalist and researcherThe movement of funds and the chances of recovery can be checked against public blockchain data.attributionRussia claims foreign intelligence services hacked officials' phonessecurity leaderState-backed targeted attacks on mobile devices bear on the protection of officials' communications.journalist and researcherState attribution should be reported as an official claim, not as fact, in the absence of independent evidence.incidentEversource says cyberattack affected 3,049 customerssecurity leaderAttacks on utilities put both customer data and operations at risk.citizenAffected customers should take the precautions set out in the notification.incidentBedfordshire Hospitals NHS Trust issues notification on Synnovis cyber incidentsecurity leaderIncidents at healthcare suppliers can affect hospitals' operations and data as well.citizenAffected patients can use the hospital's notification to check whether the incident concerns them.reportCertiK: crypto hack losses fall 90% in May to $68.3 millionsecurity leaderMonth-to-month swings in losses do not signal a lasting decline, so crypto risk still needs to be managed.journalist and researcherThe monthly loss figure is CertiK's own measurement; other data providers may report a different number.incidentCBSE says Class 12 re-evaluation portal hit by 'barrage of cyberattacks'security operationsHigh-traffic education portals are exposed to load and attack targeting around deadlines.citizenStudents and parents found it harder to complete their business through the portal because of the disruptions.influence operationMatryoshka: Russian disinformation in Armenia's electionspublic-sector decision-makerDisinformation operations during election periods require institutional preparedness.journalist and researcherThe characteristics and measured reach of the Matryoshka activity in the elections are worth documenting.attributionRussian military hacker who targeted Pennsylvania nuclear firm remains at large, report saysjournalist and researcherFurther official documents are needed on state-backed hacking aimed at critical infrastructure.law enforcementRansomware group negotiator sentenced to 8.5 years in prison over $56M cyberattackssecurity leaderThe sentence shows that the negotiator role in the ransomware ecosystem can be held legally accountable.journalist and researcherLaw enforcement cases are making the division of roles within extortion groups increasingly well documented.fraudAnother Signal phishing campaign is spreadingsecurity leaderPhishing against secure messengers also affects organisations' rules on the use of communication tools.citizenSuspicious links and code requests received on Signal should not be opened or answered.fraudFBI warns of phishing campaigns abusing the FIFA namesecurity leaderIt is worth preparing staff for campaigns around major sporting events.citizenTicket and prize offers linked to sporting events should be treated with suspicion.toolAndroid now detects deepfake calls by defaultcitizenPhone-level protection may reduce the risk of voice-cloning scams, but it does not replace caution.

← Issue 28 · Issue 30 →

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.