Artificial Intelligence
Researchers have discovered that a prompt can be disguised as an url, and accepted by Atlas as an url in the omnibox.
Hi, what are you looking for?
SecurityWeek
All posts tagged "jailbreak"
Artificial Intelligence
Researchers exploited K2 Think’s built-in explainability to dismantle its safety guardrails, raising new questions about whether transparency and security in AI can truly coexist.
Artificial Intelligence
Instead of GPT-5 Pro, your query could be quietly redirected to an older, weaker model, opening the door to jailbreaks, hallucinations, and unsafe outputs.
Artificial Intelligence
Researchers demonstrate how multi-turn “storytelling” attacks bypass prompt-level filters, exposing systemic weaknesses in GPT-5’s defenses.
Artificial Intelligence
Cisco’s latest jailbreak method reveals just how easily sensitive data can be extracted from chatbots trained on proprietary or copyrighted content.
Artificial Intelligence
The latest release of the xAI LLM, Grok-4, has already fallen to a sophisticated jailbreak.
Artificial Intelligence
New "Echo Chamber" attack bypasses advanced LLM safeguards by subtly manipulating conversational context, proving highly effective across leading AI models.
Artificial Intelligence
Cato Networks discovers a new LLM jailbreak technique that relies on creating a fictional world to bypass a model’s security controls.
Artificial Intelligence
Two Microsoft researchers have devised a new jailbreak method that bypasses the safety mechanisms of most AI systems.
Artificial Intelligence
DeepSeek’s susceptibility to jailbreaks has been compared by Cisco to other popular AI models, including from Meta, OpenAI and Google.
Artificial Intelligence
Researchers found a jailbreak method that exposed DeepSeek’s system prompt, while others have analyzed the DDoS attacks aimed at the new gen-AI.
Artificial Intelligence
Different research teams have demonstrated jailbreaks against ChatGPT, DeepSeek, and Alibaba’s Qwen AI models.
Artificial Intelligence
China’s DeepSeek blamed sign-up disruptions on a cyberattack as researchers started finding vulnerabilities in the R1 AI model.
Artificial Intelligence
Microsoft has tricked several gen-AI models into providing forbidden information using a jailbreak technique named Skeleton Key.
Cybercrime
Hackers hijacked the .gh, .sl, and .as ccTLDs and obtained HTTPS certificates for several Google domains.
Vulnerabilities
Artificial Intelligence
Cybercrime
The former core infrastructure engineer deleted admin accounts, reset hundreds of passwords, and demanded 20 bitcoin to spare the company’s servers.
Artificial Intelligence
Vulnerabilities
