Dongge Liu, Jonathan Metzman, Oliver Chang, Google Open Source Security Team Since 2016, OSS-Fuzz has been at the forefront of automated v...

Experiments with GPT-3.5 suggest that LLM-based malware reviews can complement, but not yet substitute human reviews. 1800 binary classifications performed with GPT-3.5 included false-positives and false-negatives.

Are you ready for how cybersecurity will evolve in the coming years?

Unleashing the Power of LLM and Prompt Engineering for AI Framework Fuzz

Prompt Injection Primer for Engineers. Contribute to jthack/PIPE development by creating an account on GitHub.

ShadowDragon® outlines how to identify AI generated materials online that are intentionally spreading false information or even intended to incite violence

US CISA Emphasizes Security as Foundation for AI Development

Avoiding generative AI security tools due to fear, uncertainty and doubt may be more of a risk than diving headlong into the conversation.

Even a tech pro can fall for a well-laid phishing trap. Here's what happened to me - and how you can avoid a similar fate, too.

Part 1 of a multi-part series on using LLMs securely within your organisation. This post provides a framework to categorize risks based on different use cases and deployment type.

Questions about ChatGPT-maker OpenAI's ability to comply with European privacy rules are in the frame again after a detailed complaint was filed with the Questions about ChatGPT-maker OpenAI's ability to comply with European privacy rules

Microsoft will also pay any summary judgments.

For now, SynthID works only in Google’s ecosystem — but it could someday be all over the internet

Specifically, it has been shown that LLMs can be misused for fraud, impersonation, and the generation of malware; while other authors have considered the more general problem of AI alignment.

It has been shown that even aligned language models are susceptible to adversarial attacks that bypass their restrictions on generating harmful text. We propose a simple approach to defending against these attacks by having a large language model filter its own responses.

FedMLSecurity comprises two principal components: FedMLAttacker, which simulates attacks injected into FL training, and FedMLDefender, which emulates defensive strategies designed to mitigate the impacts of the attacks.