News

Latest news on the subject of B2B cyber security >>> PR agencies: Add us to your mailing list - see contact! >>> Book an exclusive PartnerChannel for your news!

Tag: LLMs

LLMjacking: Tactics and best practices for defense
March 26, 2025
| No comments
| Stories
LLMjacking: Tactics and best practices for defense

Since its discovery in May 2024, LLMjacking has evolved along with large-scale language models (LLMs) themselves. Attackers are constantly developing new motives and methods for LLMjacking—including rapidly expanding to new LLMs like DeepSeek. As the Sysdig Threat Research Team (TRT) reported back in September, the frequency and popularity of LLMjacking attacks are increasing. Given this trend, we weren't surprised to see DeepSeek targeted within days of its media exposure and subsequent surge in usage. LLMjacking attacks have also become more prevalent in the…

Read more

DeepSeek security: Researchers test jailbreaking techniques
March 7, 2025
| No comments
| Short news
B2B Cyber ​​Security ShortNews

DeepSeek is a notable new competitor to popular AI models. LLMs with inadequate security restrictions can be used by malicious actors to create instructions for dangerous objects, such as Molotov cocktails, or even for building malware. Palo Alto Networks' Unit 42 team tested three jailbreaking techniques, "Deceptive Delight," "Bad Likert Judge," and "Crescendo," on DeepSeek models. While initial responses often seemed harmless, security researchers were able to successfully bypass the LLM's security mechanisms with follow-up prompts. DeepSeek has shared a range of malicious information, ranging from detailed instructions for creating dangerous objects to...

Read more

LLM Jacking Attacks: Alarming Increase
2 December 2024
| No comments
| Short news
B2B Cyber ​​Security ShortNews

As LLMs (Large Language Models) increase, so does the threat of LLM jacking attacks. Cybercriminals use stolen cloud login data to illegally gain access to LLMs. Companies should therefore strengthen their security measures and secure their access data. Since the discovery of LLM jacking by the Sysdig TRT, attacks have increased significantly. Attackers use stolen login information to gain access to expensive AI models and use them for their own purposes. This can cost victims enormous sums of money, as modern LLMs such as Claude 3 Opus can charge up to $100.000 per day in usage fees...

Read more

More security for generative AI and LLMs
March 11, 2024
| No comments
| PR News
More security for generative AI and LLMs

Companies cannot afford to ignore generative AI and its potential. However, you need comprehensive AI observability to ensure investments in generative AI are successful and risks are avoided. Dynatrace has expanded its analytics and automation platform to include holistic observability and security for LLMs and generative AI-powered applications. Dynatrace AI Observability enables companies worldwide to use generative AI reliably and cost-effectively to increase their innovation, productivity and sales. Major platforms supported Dynatrace AI Observability is a comprehensive solution. It covers the end-to-end AI stack, including…

Read more

© 2025 MedPressIT Selinger
Cookie Settings