Trusted AI Blog

• LLM Security

Explore the most critical vulnerabilities and emerging threats affecting Large Language Models (LLMs) and Generative AI technologies.

This week in AI security: researchers meeting in Las Vegas to work out how security vulnerabilities in AI models should be reported.

This week in AI security: the NIST generative AI profile for risk management, the world's first major AI law entering force, and mapping GenAI misuse.

This week in AI security: the launch of the Coalition for Secure AI, OpenAI closing the ignore all previous instructions loophole, and models rating safety.

This week in AI security: a technique for identifying AI bots on social media using prompt injection, and what it says about bot detection.

• LLM Security

This week in AI security: why LLM security cannot wait, ChatGPT revealing its own secret instructions, and the risk companies take on with rushed deployments.

• LLM Security

This edition covers the critical vulnerabilities and emerging threats affecting large language models and generative AI, plus CISO and VC guides.

According to CrowdStrike CTO Elia Zaitsev, the technology's capability to generate human-like text can be exploited by cybercriminals in various ways.

Generative AI, despite its potential, is susceptible to errors, biases, and poor judgment, necessitating rigorous testing methods.

• GenAI Security

This week in AI security: a framework built on data protection, identity and access management and threat monitoring, plus a course on GenAI attacks.

• GenAI Security

The 2024 AI Index report highlights a surge in AI-related patents, showing the industry's focus on innovation.

This week in AI security: a vulnerability in EmailGPT, a widely used AI email assistant, and the prompt injection risk that comes with the category.