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 33

Jun 29 – Jul 5. 2026 · 32 stories · 6 of them Hungarian and regionalPage last updated: Nov 21, 2026
Viewpoint

incident · global

Researchers say AI agent ran ransomware attack on database via Langflow flaw without human involvement

Sysdig researchers say an AI agent exploited a remote code execution (RCE) flaw in Langflow to automate a ransomware attack on a database from start to finish. Reports published on 2–3 July 2026 present it as the first such case.

2026. Jul 2. · 10 sources

Who is affected and how

security leader
Automating the attack chain can shorten the time between discovery and extortion.
security operations
Langflow instances need to be patched and isolated, and databases need strict access controls and backups.
developer
Exposed instances of AI workflow frameworks are entry points for attackers.
journalist and researcher
The researchers say the attack ran without human direction; no independent source has yet confirmed this.
vulnerability · United States

CISA: ransomware gangs now exploiting Windows BlueHammer flaw

CISA says ransomware gangs are actively exploiting a Windows flaw known as BlueHammer. The news appeared on 30 June 2026 and was reported by 14 sources.

security leaderTreat the flaw as a priority, because CISA says it is already used in ransomware attacks.security operationsPatches or mitigations must be deployed urgently across the Windows fleet, and systems should be checked for signs of compromise.public-sector decision-makerCISA's warning also bears on patching discipline in government systems that run on Windows.
vulnerability · United States

SharePoint RCE flaw (CVE-2026-45659) added to CISA KEV catalog over active exploitation

CISA has added the SharePoint remote code execution flaw CVE-2026-45659 to its catalog of known exploited vulnerabilities. The listing followed active exploitation, and the news appeared on 2 July…

security leaderThe KEV listing makes patching a deadline-bound priority, especially for internet-facing instances.security operationsSharePoint servers must be patched urgently, and systems should be checked for signs of exploitation.public-sector decision-makerInstitutions that follow federal KEV requirements may have a patching obligation.
incident · European Union

Greek politician investigating spyware vendors had phone hacked with Pegasus

A Greek politician's phone was hacked with spyware several times while they worked on a European Parliament inquiry into spyware vendors. According to one report the attack used Pegasus; the reports…

public-sector decision-makerLawmakers and committees investigating spyware cases can themselves be targets, which raises the question of protecting institutions.journalist and researcherProtecting the devices of those who investigate spyware vendors is a priority risk.citizenThe case shows that commercial spyware can also affect political oversight work.
vulnerability · global

SimpleHelp flaw added to CISA KEV catalog as attackers deploy new stealer malware

Attackers are exploiting a critical SimpleHelp flaw to deploy new stealer malware, and CISA has added the authentication bypass vulnerability to its KEV catalog. Hungary's National Cybersecurity…

security leaderBecause of the KEV listing, remote maintenance tools pose a heightened risk, especially for service providers.security operationsSimpleHelp instances must be patched or taken offline urgently, and systems should be checked for traces of past exploitation.business leaderCompanies that use remote support should check the SimpleHelp installations run by their providers.

Hungary

attributionPoland warns Russia seeks to exploit Ukraine tensions for sabotage operationssecurity leaderOrganisations in the region must reckon with the risk of physical and hybrid operations.public-sector decision-makerThe Polish assessment draws attention to the risk of sabotage against critical infrastructure in the region.fraudHungary's National Cybersecurity Institute and police warn of fraudulent messagespublic-sector decision-makerImpersonation of state bodies affects the credibility of authorities and their communication with citizens.citizenFor messages citing a fine or a state body, verify the claim through official channels instead of clicking links.toolHungary's National Cybersecurity Institute outlines Google's new AI-related data security settingcitizenThe setting may help users control data handled by AI features.fraudFraudster hunted victims on dating site, took 49 million forints from one of themcitizenOn dating sites, requests for money and pressure to act quickly can be signs of fraud.attributionHungary's National Cybersecurity Institute: Ghostwriter launches new Gmail phishing campaignsecurity leaderFor targeted phishing campaigns, strengthening user awareness and account protection is warranted.citizenSign-in prompts arriving in Gmail accounts should be treated as suspicious, and two-step verification is advisable.attributionGamaredon expands Ukraine attacks with new malware and cloud service abusesecurity operationsBecause communication is hidden in cloud services, monitoring outbound traffic is important.journalist and researcherChanges in the group's toolset should be tracked in regional threat analysis.

More key stories of the week

vulnerability · global

Cursor editor flaws let prompt injection escape the sandbox and run commands

Researchers have found sandbox-bypass flaws in the Cursor AI code editor, dubbed DuneSlide, that can let prompt injection reach the operating-system level. According to SecurityWeek a zero-click…

security leaderAI tools should be treated as an attack surface on developer endpoints.security operationsCheck that Cursor is updated and that developer machines have additional protection.developerUntrusted content processed by an AI code editor can lead to code execution on the developer's machine.journalist and researcherHere prompt injection is a demonstrated RCE vector, which is a main direction of research.
incident · United States

FBI warns of TeamPCP supply chain campaign as npm attack also targets developers

The FBI warns that the TeamPCP supply chain campaign threatens developer environments and cloud credentials. Hungary's National Cybersecurity Institute separately described an npm attack targeting…

security leaderThe FBI warning makes protecting developer environments a priority.security operationsBecause cloud credentials may be exposed, access logs should be reviewed.developerDependencies, CI environments and cloud keys should be checked, and exposed secrets rotated.
attribution

Israeli cyber chief: Iranian cyberattacks on Israel have surged since the war

The head of Israel's national cyber directorate says Iranian cyberattacks have increased since the war, targeting critical infrastructure, companies and the public. The reports cover the period after…

security leaderGiven the cyber spillover of the regional conflict, the protection of critical infrastructure and SMEs should be reviewed.public-sector decision-makerAccording to the official assessment, the cyber effects of state conflicts also reach smaller organisations and the public.
incident · Asia

iPhone 18 Pro data appears in ransomware leak; India investigates Tata breach

Alleged test images and a parts list for the iPhone 18 Pro appeared in a ransomware leak, and India is investigating the Tata data breach that exposed Apple iPhone secrets. The reports appeared…

security leaderA leak at a manufacturer in the supply chain can also expose the client's product information.business leaderManufacturing partners' data protection and incident-response capability is a business risk.
law enforcement · United States

US offers $10 million for information on group behind Signal and WhatsApp hacking spree

According to Ars Technica, the United States is offering $10 million for information on the group behind a series of attacks targeting Signal and WhatsApp accounts. GetCyberRight writes about Russian…

public-sector decision-makerOfficials who use secure messaging apps may be exposed to targeted attacks.journalist and researcherAccounts on encrypted messaging apps are targets of targeted deception.citizenTo protect Signal and WhatsApp accounts, two-step verification and checking linked devices are recommended.
incident · United States

Medtronic notifies customers of data breach linked to ShinyHunters

Medtronic has sent notifications to customers about a data breach that reports link to the ShinyHunters group. The news appeared on 2 July 2026.

security leaderData breaches at medical device makers also trigger customer notification obligations.business leaderLeaks of supplier and customer data pose regulatory and reputational risks.
incident · Asia

KDDI data breach may affect up to 14.2 million email accounts

The KDDI data breach may affect 14.2 million email accounts, according to reports. The news appeared on 29 June 2026.

security leaderExpect a wave of phishing after large email breaches at service providers.citizenPasswords of affected accounts should be changed, and phishing messages should be watched for.
tool · global

ARToken phishing panel targets Microsoft 365 accounts

A phishing panel called ARToken was built to take over Microsoft 365 accounts. The report was published on 1 July 2026.

security leaderPhishing infrastructure aimed at cloud accounts is part of everyday threats.security operationsMicrosoft 365 access policies and monitoring of anomalous sign-ins should be reviewed.
law enforcement · global

Europol-led operation dismantles networks behind three widely used hacking tools

Law enforcement agencies and private companies shut down the criminal networks behind three widely used hacking tools. News of the operation on 29 June 2026 included the SocGholish and Amadey malware…

security operationsIndicators linked to the dismantled infrastructure should be checked in network logs.journalist and researcherThe internationally coordinated operation is a further example of action against malware-as-a-service offerings.
vulnerability · global

Adobe patches seven maximum-severity flaws in ColdFusion and Campaign

Adobe has patched seven flaws rated at the highest severity in ColdFusion and Campaign products. The news was published on 1 July 2026.

security operationsAffected ColdFusion and Campaign instances must be updated with Adobe's patches.developerApplications built on ColdFusion require regression testing after the update.
law enforcement · United States

Alleged Scattered Spider member arrested in Finland and extradited to the US

The US Department of Justice announced that an alleged member of the Scattered Spider criminal group was arrested in Finland and then extradited to the United States. The announcement was published…

security leaderAction against the group shows that social engineering attacks remain a significant risk.journalist and researcherThe proceedings concern alleged membership; the outcome of the case remains open.
regulation · United States

OMB issues federal roadmap for post-quantum cryptography migration

The US Office of Management and Budget (OMB) has issued a federal roadmap for post-quantum cryptography migration. HSToday reported it on 29 June 2026.

security leaderThe cryptographic inventory and migration plan also affect government partners.public-sector decision-makerFor federal agencies, the roadmap sets out migration planning and a cryptographic inventory.

In brief

incidentNissan discloses employee data breach linked to Oracle zero-day attackssecurity operationsOracle environments should be examined for traces of zero-day exploitation.business leaderZero-day flaws in supplier software can also leak employee data.reportReport: prompt injection attacks hit more than 90 firms in 2025security leaderPrompt injection now appears in specific corporate incidents, so it should be included in the risk assessment of AI systems.developerAI applications require strict limits on inputs and outgoing actions.researchMozilla warns of indirect prompt injection risk in AI coding agentssecurity leaderAgents' permissions and approval steps need to be tightly restricted.developerExternal content read by coding agents can serve as an attack channel.incidentUS government reports another breach, with federal cloud systems affectedpublic-sector decision-makerRepeated federal incidents raise oversight questions about government cloud security.fraudWoman loses savings in deepfake romance scam involving intimate video calls with 'Dubai prince'citizenA video call is no longer proof of identity, because it can be faked with deepfakes.incidentMiddletown, Ohio, notifies more than 123,000 people of data breach involving SSNspublic-sector decision-makerMunicipal data holdings contain sensitive resident information, so protecting them is a public duty.citizenThose affected should take steps to guard against identity theft and financial fraud.incidentAflac Japan data breach exposes sensitive customer informationbusiness leaderLeaks of customer data at insurers pose regulatory and trust risks.attributionRussian hackers may have crippled Jaguar Land Rover, Hungarian report saysjournalist and researcherThe headline is in the conditional mood, so the attribution should be treated as the source's claim.incidentAmicus Solutions reports data breach involving Huntsville Hospital Health Systemsecurity leaderData breaches at hospitals' third-party providers can also affect their own patient data.

← Issue 32 · Issue 34 →

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.