Security news weekly round-up - 16th January 2026
Source: Dev.to
AI Tool Poisoning: How Hidden Instructions Threaten AI Agents
TL;DR: An AI tool can appear to work as expected while secretly stealing personal information and sending it to an attacker.
Excerpt:
Consider a scenario where an attacker publishes a tool with a seemingly harmless description. However, hidden in the metadata is an instruction to read sensitive data, such as a private key or confidential files. When the AI agent uses the tool, it unwittingly follows the malicious instruction, sharing sensitive data with the attacker. This can lead to a data breach.
Google Removes Some AI Health Summaries After Investigation Finds “Dangerous” Flaws
AI is still not perfect; it can hallucinate or generate incorrect information.
Excerpt:
The investigation revealed that searching for liver test norms generated raw data tables (listing specific enzymes like ALT, AST, and alkaline phosphatase) that lacked essential context. The AI feature also failed to adjust these figures for patient demographics such as age, sex, and ethnicity. Experts warned that because the AI model’s definition of “normal” often differed from actual medical standards, patients with serious liver conditions might mistakenly believe they are healthy.
Never‑Before‑Seen Linux Malware Is “Far More Advanced Than Typical”
Linux (and macOS) malware is becoming more visible. The article notes that the malware described is more advanced than typical, indicating a higher level of effort by its authors.
Excerpt:
VoidLink is a comprehensive ecosystem designed to maintain long‑term, stealthy access to compromised Linux systems, particularly those running on public cloud platforms and in containerized environments. Its design reflects a level of planning and investment typically associated with professional threat actors rather than opportunistic attackers, raising the stakes for defenders who may never realize their infrastructure has been quietly taken over.
Convincing LinkedIn Comment‑Reply Tactic Used in New Phishing Campaign
The phishing messages impersonate LinkedIn branding and sometimes use the official lnkd.in URL shortener, making the malicious links harder to distinguish from legitimate ones.
Excerpt:
These posts falsely claim that the user has “engaged in activities that are not in compliance” with the platform and that their account has been “temporarily restricted” until they visit the specified link in the comment.
Your Personal Information Is on the Dark Web – What Happens Next?
If your data appears on the dark web, change your login credentials immediately and consider additional protective measures.
Excerpt:
Cybercriminals primarily seek financial information (bank account numbers, card details, and logins), personally identifiable information (PII), and account credentials. With this data they can hijack accounts, drain funds, access stored card information, or conduct follow‑on phishing attacks. They may also use PII for identity fraud, such as applying for new lines of credit, medical treatment, or welfare benefits.
Cover photo by Debby Hudson on Unsplash.