New Context Compliance Attack Jailbreaks Most of The Major AI Models

Rather than requiring complex prompt engineering or computationally expensive optimization, CCA works through a basic three-step process: initiating a conversation about a sensitive topic, injecting a fabricated assistant response into the conversation history, and then having the user respond affirmatively to the fabricated question. The method works by manipulating conversation history that many AI systems rely on clients to provide, essentially tricking the AI into believing it had previously agreed to discuss harmful content. However, most open-source models and several commercial systems that depend on client-supplied conversation history remain vulnerable to this exploitation method. The technique exploits the design choice in modern AI deployment where providers do not maintain conversation state on their servers but instead rely on clients to send the full conversation history with each request. Analysts at Microsoft detected that systems which maintain conversation state on their servers—such as Copilot and ChatGPT—are not susceptible to this attack. Evaluation Table shows that models like Llama 3.1, Qwen2.5, GPT-4o, Gemini, and others are vulnerable to this attack across various sensitive content categories, while Llama2 models showed more resistance. For API-based commercial systems, potential mitigation strategies include implementing cryptographic signatures for conversation histories or maintaining limited conversation state on the server side. Unlike complex prompt engineering techniques that attempt to confuse AI systems with intricate word combinations, CCA exploits a fundamental architectural weakness present in many deployed models. The attack has successfully jailbroken numerous leading models, enabling them to generate content on sensitive topics ranging from harmful instructions to explicit material. A new, surprisingly simple method called Context Compliance Attack (CCA) has proven effective at bypassing safety guardrails in most leading AI systems. Cyber Security News is a Dedicated News Platform For Cyber News, Cyber Attack News, Hacking News & Vulnerability Analysis. These measures could help validate the integrity of conversation context and prevent the kind of manipulation that CCA exploits. With years of experience under his belt in Cyber Security, he is covering Cyber Security News, technology and other news.

This Cyber News was published on cybersecuritynews.com. Publication date: Fri, 14 Mar 2025 17:20:07 +0000


Cyber News related to New Context Compliance Attack Jailbreaks Most of The Major AI Models

Achieving Continuous Compliance - If you've ever explored regulatory compliance and cybersecurity, you'll understand the importance of continuous compliance in the digital age, where evolving technology and regulations require constant vigilance. This article will cover the ...
1 year ago Feeds.dzone.com
Leveraging Automation for Risk Compliance in IT - Organizations often encounter the challenge of managing complex technology ecosystems while ensuring data security, compliance, and risk management. One crucial aspect of this challenge is risk compliance in IT environments, specifically Linux ...
1 year ago Securityboulevard.com
Securing AI: Navigating the Complex Landscape of Models, Fine-Tuning, and RAG - It underscores the urgent need for robust security measures and proper monitoring in developing, fine-tuning, and deploying AI models. The emergence of advanced models, like Generative Pre-trained Transformer 4, marks a new era in the AI landscape. ...
1 year ago Feedpress.me
How to Get PCI Compliance Certification? Steps to Obtain it - To mitigate the risk of such breaches, PCI compliance establishes stringent security protocols. In this blog let's understand how to get PCI Compliance certification. PCI DSS is a security standard for card transactions, which includes detailed ...
10 months ago Securityboulevard.com
New Context Compliance Attack Jailbreaks Most of The Major AI Models - Rather than requiring complex prompt engineering or computationally expensive optimization, CCA works through a basic three-step process: initiating a conversation about a sensitive topic, injecting a fabricated assistant response into the ...
3 hours ago Cybersecuritynews.com
Enhancing PCI DSS Compliance: The Urgent Need for Risk-Based Prioritization - Keeping U.S. commercial critical national infrastructure organizations safe is vital to national security, and it's never been more top of mind as international conflicts and cyberattacks increase and create tensions for businesses, governments, and ...
1 year ago Cyberdefensemagazine.com
A New Trick Uses AI to Jailbreak AI Models-Including GPT-4 - Large language models recently emerged as a powerful and transformative new kind of technology. Their potential became headline news as ordinary people were dazzled by the capabilities of OpenAI's ChatGPT, released just a year ago. In the months that ...
1 year ago Wired.com
Legal and Compliance Considerations in Cloud Computing - This paradigm change has faced challenges, primarily legal and compliance issues. This can present severe legal issues, particularly regarding data ownership. According to S. Krishnan, the transforming nature of computing has created legal ...
1 year ago Feeds.dzone.com
Coming March 2024: How to Prepare for PCI DSS Version 4.0 Compliance - A 2022 Verizon report claims that only 43% of assessed organizations maintained full compliance in 2020. With the March 2024 deadline fast approaching, businesses that process and store card data are racing to implement the 13 new requirements in ...
1 year ago Securityboulevard.com
In the rush to build AI apps, don't leave security behind The Register - There are countless models, libraries, algorithms, pre-built tools, and packages to play with, and progress is relentless. You'll typically glue together libraries, packages, training data, models, and custom source code to perform inference tasks. ...
11 months ago Go.theregister.com Hunters
How machine learning helps us hunt threats | Securelist - In this post, we will share our experience hunting for new threats by processing Kaspersky Security Network (KSN) global threat data with ML tools to identify subtle new Indicators of Compromise (IoCs). The model can process and learn from millions ...
5 months ago Securelist.com
Achieving Automated TISAX Compliance - In its 2024 Automotive Cybersecurity Report, Upstream found that 50% of all automotive cyber incidents in 2023 had a high or massive impact. International institutions are taking steps to help automotive organizations defend themselves against black ...
9 months ago Tripwire.com
Addressing Deceptive AI: OpenAI Rival Anthropic Uncovers Difficulties in Correction - There is a possibility that artificial intelligence models can be trained to deceive. According to a new research led by Google-backed AI startup Anthropic, if a model exhibits deceptive behaviour, standard techniques cannot remove the deception and ...
1 year ago Cysecurity.news
5 Unique Challenges for AI in Cybersecurity - Applied AI in cybersecurity has many unique challenges, and we will take a look into a few of them that we are considering the most important. On the other hand, supervised learning systems can remediate this issue and filter out anomalous by design ...
1 year ago Paloaltonetworks.com
Optimize Control Health Management Across Business Levels: Introducing Scopes - Managing controls across multiple business units becomes increasingly challenging and costly as operational requirements evolve. To help compliance leaders efficiently view and manage control health across product lines, geographies and business ...
1 year ago Securityboulevard.com
Localization Mandates, AI Regs to Pose Major Data Challenges in 2024 - Companies should expect to face a trio of trends in 2024 that make data security, protection, and compliance more critical to operations and risk reduction. Increasingly, governments worldwide are creating laws that govern the handling of data within ...
1 year ago Darkreading.com
9 Best DDoS Protection Service Providers for 2024 - eSecurity Planet content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More. One of the most powerful defenses an organization can employ against distributed ...
1 year ago Esecurityplanet.com
Attack Vector vs Attack Surface: The Subtle Difference - Cybersecurity discussions about "Attack vectors" and "Attack surfaces" sometimes use these two terms interchangeably. This article guides you through the distinctions between attack vectors and attack surfaces to help you better understand the two ...
2 years ago Trendmicro.com
ML Model Repositories: The Next Big Supply Chain Attack Target - The techniques are similar to ones that attackers have successfully used for years to upload malware to open source code repositories, and highlight the need for organizations to implement controls for thoroughly inspecting ML models before use. ...
11 months ago Darkreading.com
Master Security by Building on Compliance with A Risk-Centric Approach - In recent years, a confluence of circumstances has led to a sharp rise in IT risk for many organizations. That's why a proactive approach to seeing, understanding, and acting on risk is key to improving the effectiveness of defenses in place to meet ...
1 year ago Cyberdefensemagazine.com
Startups Scramble to Build Immediate AI Security - It also elevated startups working on machine learning security operations, AppSec remediation, and adding privacy to AI with fully homomorphic encryption. AI's largest attack surface involves its foundational models, such as Meta's Llama, or those ...
1 year ago Darkreading.com
Business Data Privacy Laws: Compliance and Beyond - Governments worldwide have implemented strict data privacy laws to protect individuals' information in the face of increasing cyber threats and data breaches. Let's dive into the world of business data privacy laws as we navigate the complexities of ...
1 year ago Securityzap.com
Protect AI Unveils Gateway to Secure AI Models - Protect AI today launched a Guardian gateway that enables organizations to enforce security policies to prevent malicious code from executing within an artificial intelligence model. Guardian is based on ModelScan, an open source tool from Protect AI ...
1 year ago Securityboulevard.com
Cybercriminals are Showing Hesitation to Utilize AI Cyber Attacks - Media reports highlight the sale of LLMs like WormGPT and FraudGPT on underground forums. Fears mount over their potential for creating mutating malware, fueling a craze in the cybercriminal underground. Concerns arise over the dual-use nature of ...
1 year ago Cybersecuritynews.com

Cyber Trends (last 7 days)