Blog

LLM security advice looks solid until you check the hard cases

Plenty of people now type their security worries straight into a chatbot. A hacked account, a suspicious email, a stalker who might be tracking a phone, all of it lands in the same window someone would use to ask about dinner. A benchmark called HelpBench tests how well chatbots handle those moments, and the results give security professionals something to watch in what their users are being told. Researchers at University College London and Google … More → The post LLM security advice looks solid until you check the hard cases appeared first on Help Net Security.
http://news.poseidon-us.com/TTCGC3

Scoring AI hackers when there is no answer key

AI models are solving more and more of the offensive-cyber tests built to measure them. Once a model solves most of a benchmark, that benchmark runs out of room and says little about the best systems anymore. Many of those tests also lean on bugs that already have public writeups, so a strong score can come partly from a model repeating something it has read. FrontierCyber, a benchmark from the AI security lab Irregular, goes … More → The post Scoring AI hackers when there is no answer key appeared first on Help Net Security.
http://news.poseidon-us.com/TTCGB4

Why AI governance matters at scale

Your next AI agent may need to be treated like a new employee. 
http://news.poseidon-us.com/TTCBnR

Best practices for AI in open-source work

Free and open source software developers us AI coding assistants such as Claude Code, Copilot CLI, Antigravity, and OpenCode in their daily work. The Software Freedom Conservancy responded to that trend with a set of recommendations for contributors who use these tools, which it groups under the label LLM-gen-AI, meaning generative AI systems backed by LLMs. Best practices, ordered by priority The document contains 14 recommendations presented in the authors’ order of importance. The Conservancy … More → The post Best practices for AI in open-source work appeared first on Help Net Security.
http://news.poseidon-us.com/TTCBd6

What your next cyber insurance renewal will demand

In this Help Net Security video, Michael Loewy, co-founder, Tide Foundation, explains how cyber insurance is rewriting security programs at renewal time. Insurers want more questionnaires, more evidence, and more attestations, because the market is moving from trusting your answers to verifying them. Loewy argues that compliance is not security, and that a control which is declared can be partial, misconfigured, or bypassed when it matters. He walks through where proof beats paperwork: MFA that … More → The post What your next cyber insurance renewal will demand appeared first on Help Net Security.
http://news.poseidon-us.com/TTCBd0

Most teams will ship AI-written infrastructure code with little review

AI-assisted development has settled into everyday practice across software organizations, and developers using it move from idea to working code in hours. That code does not stay with the developers who prompt it. It flows downstream to the DevOps and platform teams who deploy and maintain it, and those teams are not getting the same speed boost. A Spacelift survey of 406 IT and platform leaders in North America captures what happens when one side … More → The post Most teams will ship AI-written infrastructure code with little review appeared first on Help Net Security.
http://news.poseidon-us.com/TTCBcB