AbejaIT AbejaIT
AbejaIT

Blog

Technology news, cybersecurity, AI and IT infrastructure — our articles and carefully curated industry updates.

We publish our own analysis and practical insights from AbejaIT projects, and we also aggregate and summarize valuable content from trusted industry sources. You will find updates on software, cloud, security, artificial intelligence and IT trends — written for both business decision-makers and technical teams.

Stop Your Legacy Infrastructure from Hijacking Your AI Agents

Stop Your Legacy Infrastructure from Hijacking Your AI Agents

Legacy IT infrastructure is becoming an increasingly serious threat in the era of AI agents — attackers actively exploit legacy systems as entry points to hijack control over autonomous AI-powered processes. Despite over 70% of organizations running AI agent pilots, security programs often fail to keep pace with this adoption rate. Infrastructure modernization is no longer optional — it is the foundation of a secure AI strategy.

Read more
Google Sets Sept. 30 Deadline for Android Developer Verification in Four Countries

Google Sets Sept. 30 Deadline for Android Developer Verification in Four Countries

Google announced that from September 30, 2026, certified Android devices in Brazil, Indonesia, Singapore, and Thailand will block app installations from developers who have not verified their identity. This is the first phase of a global rollout of mandatory developer verification aimed at limiting malware distribution. Companies distributing their own mobile apps should check the registration status of their developer accounts.

Read more
New OXLOADER Loader Uses Malicious Google Ads to Deliver CastleStealer

New OXLOADER Loader Uses Malicious Google Ads to Deliver CastleStealer

Cybersecurity researchers discovered a new malware campaign using a previously unknown loader called OXLOADER to distribute CastleStealer. Attackers use malicious Google Ads as an infection vector, making the threat particularly difficult to detect. Analysts at Elastic Security Labs indicate the campaign is likely operated by Russian-speaking financially motivated cybercriminals.

Read more
Researchers Detail DifyTap Flaws in Dify That Could Expose AI Chats Across Tenants

Researchers Detail DifyTap Flaws in Dify That Could Expose AI Chats Across Tenants

Researchers from Zafran Security disclosed four vulnerabilities in Dify — a popular open-source platform for building agentic AI workflows — collectively named DifyTap. The flaws enable unauthorized access to AI model conversations belonging to other tenants, posing a serious threat in multi-tenant environments. Organizations using Dify should prioritize verifying their configuration and monitoring for updates.

Read more
29-Year-Old Squid Proxy Bug 'Squidbleed' Can Leak Cleartext HTTP Requests

29-Year-Old Squid Proxy Bug 'Squidbleed' Can Leak Cleartext HTTP Requests

A critical vulnerability in the popular Squid proxy, dubbed 'Squidbleed,' was discovered after remaining unnoticed for nearly three decades in code dating back to 1997. The heap over-read bug enables leaking cleartext HTTP requests from other users — including authentication credentials and session tokens. Organizations using Squid should immediately verify their software version and apply available patches.

Read more
ShapedPlugin WordPress Pro Plugins Backdoored in Supply Chain Attack

ShapedPlugin WordPress Pro Plugins Backdoored in Supply Chain Attack

ShapedPlugin's WordPress plugins fell victim to a supply chain attack — hackers took control of official distribution channels and injected malicious backdoor code into Pro versions. WordPress administrators should immediately verify installed ShapedPlugin plugins and conduct a security audit of their websites.

Read more
OpenAI Expands Daybreak With GPT-5.5-Cyber to Help Defenders Patch Security Flaws

OpenAI Expands Daybreak With GPT-5.5-Cyber to Help Defenders Patch Security Flaws

OpenAI is expanding its Daybreak program, making the enhanced GPT-5.5-Cyber model available to trusted security professionals — according to the company, its most effective AI tool yet for detecting and patching software vulnerabilities. The model's ability to deeply analyze large codebases can significantly accelerate the work of defensive security teams.

Read more
WhatsApp VBScript Campaign Uses Fake Documents to Install ManageEngine RMM Tool

WhatsApp VBScript Campaign Uses Fake Documents to Install ManageEngine RMM Tool

Kaspersky detected an active cyberattack campaign in which criminals use WhatsApp to distribute malicious VBScript files disguised as documents. Once executed, the script installs a legitimate RMM tool (ManageEngine), giving attackers remote control over the victim's device. The campaign targets users in multiple countries including the UK, India, Brazil, and Singapore.

Read more
Malicious npm Packages Pose as PostCSS Tools to Deliver Windows RAT

Malicious npm Packages Pose as PostCSS Tools to Deliver Windows RAT

Cybersecurity researchers discovered malicious npm packages masquerading as PostCSS tools that actually install a Remote Access Trojan (RAT) on Windows systems. The packages accumulated over a thousand downloads before being identified. This is another example of a software supply chain attack — developers should verify the source and authenticity of every dependency before deploying it in a project.

Read more
Technologies and partners we work with