Close Menu
Indo Guard OnlineIndo Guard Online
  • Home
  • Cyber Security
  • Risk Management
  • Travel
  • Security News
  • Tech
  • More
    • Data Privacy
    • Data Protection
    • Global Security
What's Hot

Invitation to Disagreement Link from ASYNCRAT and SKULD Theft, focused on cry

June 14, 2025

More than 269 000 sites infected with malicious JSFiretruC JavaScript software in one month

June 13, 2025

Transition from Monitoring Alert to Risk Measurement

June 13, 2025
Facebook X (Twitter) Instagram
Facebook X (Twitter) Instagram YouTube
Indo Guard OnlineIndo Guard Online
Subscribe
  • Home
  • Cyber Security
  • Risk Management
  • Travel
  • Security News
  • Tech
  • More
    • Data Privacy
    • Data Protection
    • Global Security
Indo Guard OnlineIndo Guard Online
Home » The new “file rules, attack” allows hackers to enter malicious code through the AI ​​code editors
Global Security

The new “file rules, attack” allows hackers to enter malicious code through the AI ​​code editors

AdminBy AdminMarch 18, 2025No Comments2 Mins Read
Malicious Code via AI Code Editors
Share
Facebook Twitter LinkedIn Pinterest Email Copy Link


March 18, 2025Red LakshmananAI / safety software safety

Malicious code through the editors of AI code

Cybersecurity researchers have revealed details about the new vector of the supply chain called “sourdough”, which affects artificial intelligence (AI), which is supported by code editor such as GitHub Copilot and Cersor, which will introduce a malicious code.

“This technique allows hackers silently endangered AI-Zeented code by introducing hidden malicious instructions into the seemingly innocent file – Note In a technical report that is shared with Hacker News.

Cybersecurity

“Using hidden Unicode characters and sophisticated evading methods in the model facing useful load, threatening subjects can manipulate AI to insert a malicious code that bypasses typical code reviews.”

The attack vector is characterized by the fact that it allows the malicious code to spread silently through projects, creating the risk of supply chain.

Malicious code through the editors of AI code

The essence of the attack depends on File rules AI -used AI agents to guide their behavior, helping users identify the best project coding practices and architecture.

In particular, it involves the installation of carefully designed tips in seemingly benign rules files, causing the AI ​​tool to create a code that contains safety or rear rooms. In other words, poisoned rules push the II to produce a moody code.

This can be achieved using the shirts, two-directed text markers and other invisible characters to hide malicious instructions and use the II’s ability to interpret the natural language to create a vulnerable code using semantic patterns that cheat the model in reassessment of ethical and safe restrictions.

Cybersecurity

Following the responsible disclosure of information at the end of February and March 2024, both Cersor, and GiHub, said that users were responsible for the consideration and acceptance of the offers created by the instruments.

“” Rules File “behind is a significant risk by armed by AI as a vector of the attack, effectively turning the most trusted assistant developer into a muzzle complicit, which potentially affects millions of end users through the broken software,” Carliner said.

“Once the file of the poisoned rules are included in the project repository, it affects all future sessions of the Code Code.

Found this article interesting? Keep track of us further Youter  and LinkedIn To read more exclusive content we publish.





Source link

Share. Facebook Twitter Pinterest LinkedIn Tumblr Email Copy Link
Admin
  • Website

Related Posts

Invitation to Disagreement Link from ASYNCRAT and SKULD Theft, focused on cry

June 14, 2025

More than 269 000 sites infected with malicious JSFiretruC JavaScript software in one month

June 13, 2025

Transition from Monitoring Alert to Risk Measurement

June 13, 2025

Band

June 13, 2025

Apple Zero Click’s downside in reports to spy on journalists using spyware Paragon software

June 13, 2025

Both Vextrio and affiliates control the global network

June 12, 2025
Add A Comment
Leave A Reply Cancel Reply

Loading poll ...
Coming Soon
Do You Like Our Website
: {{ tsp_total }}

Subscribe to Updates

Get the latest security news from Indoguardonline.com

Latest Posts

Invitation to Disagreement Link from ASYNCRAT and SKULD Theft, focused on cry

June 14, 2025

More than 269 000 sites infected with malicious JSFiretruC JavaScript software in one month

June 13, 2025

Transition from Monitoring Alert to Risk Measurement

June 13, 2025

Band

June 13, 2025

Apple Zero Click’s downside in reports to spy on journalists using spyware Paragon software

June 13, 2025

Both Vextrio and affiliates control the global network

June 12, 2025

How to Decide Safety Expanding

June 12, 2025

The new tokenbreak attack combines AI moderation with a one -sided character change

June 12, 2025
About Us
About Us

Provide a constantly updating feed of the latest security news and developments specific to Indonesia.

Facebook X (Twitter) Pinterest YouTube WhatsApp
Our Picks

Invitation to Disagreement Link from ASYNCRAT and SKULD Theft, focused on cry

June 14, 2025

More than 269 000 sites infected with malicious JSFiretruC JavaScript software in one month

June 13, 2025

Transition from Monitoring Alert to Risk Measurement

June 13, 2025
Most Popular

In Indonesia, crippling immigration ransomware breach sparks privacy crisis

July 6, 2024

Why Indonesia’s Data Breach Crisis Calls for Better Security

July 6, 2024

Indonesia’s plan to integrate 27,000 govt apps in one platform welcomed but data security concerns linger

July 6, 2024
© 2025 indoguardonline.com
  • Home
  • About us
  • Contact us
  • Privacy Policy

Type above and press Enter to search. Press Esc to cancel.