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 6

Dec 22–28. 2026 · 33 stories · 4 of them Hungarian and regionalPage last updated: Nov 21, 2026
Viewpoint

vulnerability · global

WatchGuard Firebox flaw CVE-2025-14733 under active attack, over 115,000 devices may be vulnerable

WatchGuard has patched a remote code execution flaw in Firebox firewalls, tracked as CVE-2025-14733, which attackers reportedly exploited. According to Help Net Security, more than 115,000 Firebox devices may be vulnerable (22 December 2025).

2025. Dec 22. · 6 sources

Who is affected and how

security leader
An actively exploited, zero-day-type flaw on an edge device requires an immediate risk assessment.
security operations
The patch must be installed urgently on Firebox firewalls, and systems should be checked for signs of earlier compromise.
incident · European Union

DDoS attack before Christmas disrupts French postal and banking services

La Poste, the French postal service, saw its services disrupted on Monday 22 December 2025 amid suspected cyberattack, with parcel delivery delayed and online payments affected. According to…

security leaderA DDoS attack carried out during the holiday peak hits the availability of parcel and payment services.security operationsProtection against volumetric attacks and a review of capacity reserves are due ahead of the holiday season.citizenLate deliveries and unavailable online payments directly affected French customers.
report · global

Crypto theft hits record in 2025; one report attributes $2.02 billion to North Korean hackers

A report of 22 December 2025 says North Korea-linked hackers stole $2.02 billion in 2025, while another article reports a record $2.7 billion in annual crypto theft following major exchange hacks…

security leaderAt crypto service providers and exchanges, key and access management targeted by state-level attackers is the main risk.business leaderFor companies holding crypto assets, record annual losses warrant a review of the insurance and custody model.journalist and researcherBecause the two totals differ, check the source's methodology before adopting a figure.
report · global

OpenAI says AI browsers may always be vulnerable to prompt injection attacks

OpenAI says AI browsers may remain vulnerable to prompt injection attacks indefinitely. At the same time, the company published an article on continuously hardening ChatGPT Atlas (22 December 2025).

security leaderWhen deploying agent-capable browsers, prompt injection should be treated as a residual risk, not as a threat that can be fully ruled out.developerWhen designing agentic features, narrowing permissions and treating external content as untrusted are essential.business leaderDecisions on enterprise use of AI browsers should take into account the vendor's own acknowledgement of the risk.
vulnerability · global

'LangGrinch' vulnerability in langchain-core puts AI agent secrets at risk

A critical vulnerability in langchain-core, dubbed "LangGrinch", puts the secrets of AI agents at risk, SiliconANGLE reported. The story appeared on 25 December 2025.

security operationsTracking the dependencies of agent frameworks and rotating exposed credentials is advisable.developerDevelopers of agents built on langchain-core need to update and review the secrets their agents can access.

Hungary

More key stories of the week

vulnerability · global

MongoBleed vulnerability (CVE-2025-14847) is being exploited by attackers

According to the report, the flaw known as MongoBleed, CVE-2025-14847, is being actively exploited. The article was published on 28 December 2025.

security operationsPatching MongoDB instances and assessing network exposure is urgent.developerOperators of applications built on MongoDB need to check the versions in use.
law enforcement

Interpol's Operation Sentinel: 574 arrests in Africa, decryption of six ransomware families

Operation Sentinel, coordinated by Interpol, led to 574 arrests and the recovery of $3 million, in connection with business email compromise. Reports say decryption was also achieved for six…

security leaderBusiness email compromise remains a financially damaging form of attack that is a continuing focus of law enforcement.security operationsVictims of the decrypted ransomware families can check whether they were affected and whether a decryptor is available.citizenThrough international cooperation, victims may receive help in recovering their data.
incident · global

Trust Wallet browser extension hack drains at least $6–7 million in crypto

An exploit of Trust Wallet's version 2.68 Chrome extension reportedly drained more than $6 million from users, while other reports put the figure at $7 million. About $4 million of the stolen funds…

security operationsBecause the stolen funds are flowing to exchanges, tracing is important for exchange compliance teams.developerThe release and update process for browser extensions carries supply-chain risk, so verifying releases is a key issue.citizenUsers of the Trust Wallet extension should check the extension version and their wallet activity.
incident · United States

University of Phoenix data breach affects nearly 3.5 million people

The University of Phoenix data breach affects nearly 3.5 million people, according to the disclosure. The news appeared on 22 December 2025.

security leaderLarge stores of personal data at educational institutions are prime targets, so a reassessment of data-handling and access controls is warranted.citizenAffected individuals' personal data may have fallen into unauthorised hands, so it is worth watching for later fraud attempts.
incident · global

Nissan says thousands of customers exposed in Red Hat data breach

Nissan says data of thousands of customers became accessible in a breach affecting Red Hat. The news was reported on 22 December 2025.

security leaderAn incident at a supplier or service provider can also affect a company's own customer data, so third parties' data handling must be checked.citizenAffected Nissan customers should watch for notifications and expect targeted phishing messages.
vulnerability · United States

CISA adds Digiever DS-2105 Pro NVR command injection flaw to KEV catalog

CISA has added a critical command injection flaw in the Digiever DS-2105 Pro network video recorder to its Known Exploited Vulnerabilities catalog. The move was reported on 22 December 2025.

security leaderThe KEV listing indicates active exploitation, so inventorying and segmenting camera systems is a priority.security operationsDigiever NVR devices affected by the KEV-listed flaw must be patched or isolated from the internet.
incident · United States

Aflac's June data breach affects more than 22 million customers

Aflac's June data breach affects more than 22 million customers, according to a report of 23 December 2025. The source summary gives no further details.

security leaderFor insurers, protecting large volumes of customer data and timing the notification process are priority issues.citizenThose affected should prepare for future misuse of their personal data, such as targeted fraud.
report · global

Ransomware.live: 306 extortion groups and 7,902 victims in 2025

According to data from the Ransomware.live tracker, 306 groups were active over the past year, and 7,902 victims were listed at the time of writing. Infosecurity Magazine summarised the trends on 23…

security leaderVictim-list data show the scale of the ransomware threat, but they rely on the groups' own disclosures.journalist and researcherThe figures are based on Ransomware.live tracking and have methodological limitations.
regulation · United Kingdom

UK Cyber Security and Resilience Bill: what the proposed law means

The article discusses what the UK's Cyber Security and Resilience Bill means. It was published on 22 December 2025.

security leaderThe UK regulatory direction may affect the obligations of service providers and the sectors concerned.public-sector decision-makerThe UK proposal offers a point of comparison for EU NIS2 practice.
research · global

AI safety dialogs can be exploited for remote code execution

The Hacker News reports that AI safety dialogs can be exploited for remote code execution. The excerpt gives no further detail (22 December 2025).

security operationsChecking the approval mechanisms of AI development tools and agents is warranted.developerApproval and safety dialogs in AI tools can become an attack surface, so how they are handled needs to be planned.
report · global

Estimate: 95% of companies do not protect autonomous agents with identity security

According to CSO Online, an estimated 95% of companies have not introduced identity protection for their autonomous agents. The article was published on 23 December 2025.

security leaderIdentity and access management for agents is, by this estimate, still missing at most organisations.business leaderEstablishing identity-protection basics before deploying agents is a risk-reducing step.
research · global

WebRAT malware spread via fake vulnerability exploits on GitHub

WebRAT malware is being distributed on GitHub disguised as vulnerability exploits, according to reports. The news dates from 23 December 2025.

security operationsExploit code downloaded from GitHub should be treated as potentially infected.journalist and researcherFake proof-of-concept exploits are a risk for security researchers, so they should be tested only in an isolated environment.

In brief

incidentCoupang says all leaked customer information has been deletedsecurity leaderThe company's own statement that the data was deleted can be given only limited weight without independent verification.citizenAffected customers should remain cautious despite the statement that the data has been deleted.law enforcementCoinbase says former agent arrested in India over exchange hacksecurity leaderAccess rights and data access held by insiders or external staff at service providers are a key risk.business leaderVetting outsourced staff who have access to customer data is a management responsibility.incidentUK minister confirms government was hacked in Octobersecurity leaderThe source summary gives no details, so following official statements is necessary.public-sector decision-makerPublic acknowledgement of a successful attack on government systems offers a basis for comparing incident-handling and communication practices.regulationEU sanctions Diana Panchenko over foreign information manipulation (FIMI)public-sector decision-makerThe EU sanctions toolkit can also be used against information manipulation activities.journalist and researcherThe sanction shows the EU's FIMI concept being applied in practice, without any measurement of its impact.attributionNorth Korea-backed hackers use new file-based attack method, Yonhap reportssecurity operationsBecause delivery is file-based, analysing attachments and documents remains a key point of defence.journalist and researcherIn the absence of technical details, it is worth tracking down the original source of the report.law enforcementNefilim ransomware affiliate pleads guilty to computer fraudsecurity leaderSuccessful prosecutions of ransomware actors are examples of legal deterrence, but they do not replace defensive work.journalist and researcherThe case is a documented instance of holding the ransomware ecosystem to account.incidentKoinly reports email leak after breach at third-party vendor Mixpanelsecurity leaderCustomer data held by third-party analytics providers falls within a company's own risk perimeter.citizenKoinly users should expect targeted phishing emails.incidentNS Power releases redacted report on 2025 cyberattacksecurity leaderSimilar utilities should review the lessons of an energy provider's incident report.public-sector decision-makerPublic accountability for incidents affecting critical infrastructure is a regulatory question.vulnerabilityMarshmallow Schema.load(many) denial-of-service flaw (CVE-2025-68480)developerApplications that use Marshmallow should switch to the patched version and limit input data.attributionIranian-linked Handala group claims to have hacked Netanyahu aide's phonesecurity leaderMobile devices of senior officials are prime targets and need a separate level of protection.journalist and researcherThe group's own claim cannot be treated as fact until independently confirmed.law enforcement54 people charged in US over ATM attacks using Ploutus malwaresecurity leaderBanks and ATM operators need to handle the physical and software protection of their machines together.security operationsDetecting Ploutus-like malware and logging physical access to ATMs are the top priorities.incidentHungarian business portal reports huge data breach at SpotifycitizenSpotify users should follow official announcements about their data.

← Issue 5 · Issue 7 →

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.