"I Had a Dream" and Generative AI Jailbreaks

A message from ChatGPT followed by a piece of malicious code and a brief remark not to use it for illegal purposes. Initially published by Moonlock Lab, the screenshots of ChatGPT writing code for a keylogger malware is yet another example of trivial ways to hack large language models and exploit them against their policy of use.

The case with the dream is just one of many jailbreaks actively used to bypass content filters of the generative AI. Even though each LLM introduces moderation tools that limit their misuse, carefully crafted reprompts can help hack the model not with strings of code but with the power of words.

Posted from: Click here for the full article.

留言

發佈留言

此網誌的熱門文章

Microsoft Expands Free Logging Capabilities for all U.S. Federal Agencies

Microsoft has expanded free logging capabilities to all U.S. federal agencies using Microsoft Purview Audit irrespective of the license tier, more than six months after a China-linked cyber espionage campaign targeting two dozen organizations came to light. "Microsoft will automatically enable the logs in customer accounts and increase the default log retention period from 90 days to 180 days," Posted from: Click here for the full article .
閱讀更多

Zimbra Warns of Critical Zero-Day Flaw in Email Software Amid Active Exploitation

Zimbra has warned of a critical zero-day security flaw in its email software that has come under active exploitation in the wild. "A security vulnerability in Zimbra Collaboration Suite Version 8.8.15 that could potentially impact the confidentiality and integrity of your data has surfaced," the company said in an advisory. It also said that the issue has been addressed and that it's expected to Posted from: Click here for the full article .
閱讀更多

New Chrome Vulnerability Enables Cross-Origin Data Leak via Loader Referrer Policy

Google on Wednesday released updates to address four security issues in its Chrome web browser, including one for which it said there exists an exploit in the wild. The high-severity vulnerability, tracked as CVE-2025-4664 (CVSS score: 4.3), has been characterized as a case of insufficient policy enforcement in a component called Loader. "Insufficient policy enforcement in Loader in Google Posted from: Click here for the full article .
閱讀更多