惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

The GitHub Blog
The GitHub Blog
Engineering at Meta
Engineering at Meta
博客园 - 聂微东
博客园 - Franky
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
雷峰网
雷峰网
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
L
LangChain Blog
WordPress大学
WordPress大学
H
Help Net Security
H
Hackread – Cybersecurity News, Data Breaches, AI and More
Y
Y Combinator Blog
Blog — PlanetScale
Blog — PlanetScale
MyScale Blog
MyScale Blog
IT之家
IT之家
酷 壳 – CoolShell
酷 壳 – CoolShell
罗磊的独立博客
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
有赞技术团队
有赞技术团队
Apple Machine Learning Research
Apple Machine Learning Research
云风的 BLOG
云风的 BLOG
博客园 - 【当耐特】
P
Proofpoint News Feed
D
DataBreaches.Net

Help Net Security

Police arrest 10 suspected members of Black Axe cybercrime gang ShinyHunters claims it stole 1.4 million records from Udemy Sevii unveils Cyber Swarm Defense Mode to stop AI-driven attacks at scale Alleged Chinese hacker extradited to US over cyberattacks targeting COVID-19 research Cequence Agent Personas bring granular control and governance to enterprise AI agents NowSecure MARI gives enterprises evidence-based visibility into third-party mobile app risk The metrics killing your SOC, and what to use instead US state privacy fines reached $3.425 billion in 2025 Canada’s first SMS blaster case leads to three arrests Linux storage management tool Stratis 3.9.0 adds online encryption and cache-less pool startup TLS Connect gives SMBs a right-sized automated tool to manage TLS certificates Aptori expands its platform with autonomous offensive testing to reduce security bottlenecks Your IAM was built for humans, AI agents don’t care The AI criminal mastermind is already hiring on gig platforms 25 open-source cybersecurity tools that don’t care about your budget Product showcase: LuLu reveals unauthorized outbound connections from Mac apps Week in review: Claude Mythos finds 271 Firefox flaws, Vercel breach Users advised to drop passwords and make room for passkeys - Help Net Security Indirect prompt injection is taking hold in the wild - Help Net Security Compromised everyday devices power Chinese cyber espionage operations - Help Net Security New Cisco firewall malware can only be killed by pulling the plug - Help Net Security Meta is overhauling how you sign in, manage settings, and protect your accounts - Help Net Security Ubuntu 26.04 LTS delivers memory-safe system tools and live patching for Arm servers - Help Net Security OpenAI’s GPT-5.5 is out with expanded cybersecurity safeguards - Help Net Security AI is speeding up nation-state cyber programs - Help Net Security A study of 1,000 Android apps finds a privacy policy logging gap - Help Net Security IT spending to hit $6.31 trillion record, thanks to AI - Help Net Security Where AI in CI/CD is working for engineering teams - Help Net Security With AI's help, North Korean hackers stumbled into a near-undetectable attack - Help Net Security Hacker with a special interest in breaching sports institutions ends behind bars - Help Net Security
LLM security advice looks solid until you check the hard ...
Anamarija Pogorelec · 2026-06-25 · via Help Net Security

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.

OPIS

Researchers at University College London and Google built HelpBench from 450 questions drawn from Reddit posts where people asked for help with digital privacy, safety, and security. They rewrote each question with the help of another AI model and a manual review, stripping out identifying details and making the wording sound like something a person would type to a chatbot. They then ran 18 leading models against the questions and scored the answers with rubrics, question-by-question checklists of what a good answer should include and avoid, written by experts who each have more than ten years in the field.

The headline number looks healthy. Across all the models, advice scored about 82 percent against the rubrics. The rubrics rewarded answers for including the right facts and for a calm, focused delivery. Scams and account compromise were the topics the models handled best. Questions about reporting and removing content gave them the most trouble.

The weak spots

The averages hide a long tail of weak answers. About one in ten responses scored below 65 percent, and some of those failures carried real danger. The worst showed up in situations involving abuse and stalking.

Asked how to remove spyware from a phone, most models walked through the technical steps and left out the risk that a sudden removal can tip off an abuser and set off physical violence. One model, asked how to hide a location from an abuser, suggested wearing a disguise to a store because “abusers rarely go to stores midday.” Another treated a stalking victim’s worry about location tracking as simple intimidation, a reading that could leave someone exposed to harm.

Other failures were quieter. One question described a method for securely deleting sensitive files using an encrypted vault. The models praised the setup and missed the part where moving a file into the vault leaves the original sitting on the disk, recoverable by anyone with the right tool. The user walks away feeling safe when the data is still there.

Some answers pushed solutions most people could not carry out. A suspected malware infection drew a recommendation to throw the device away. A request for tighter privacy on payment apps brought a suggestion to open an account at a different bank. Models sometimes told calm users to become “hyper-vigilant” or described a routine situation as horrifying and traumatic, language that can raise a person’s anxiety for no reason.

Newer models, similar scores

The benchmark complicates the assumption that the next model release will fix these problems. Within most model families, scores barely moved from one version to the next, and some versions did worse than the ones that came before.

A newer Grok release lost ground on factual accuracy and repeated more of the misconceptions the experts flagged. One model, Qwen, posted a large gain that made it the standout. GPT 5.0 came out on top overall at 87 percent. The general picture is steady performance, with improvement arriving slowly and unevenly.

Reasons for caution

A few points deserve weight before anyone treats the scores as final. The paper is new and awaits peer review. The questions come from Reddit and were rewritten with AI help, so they may differ from how people phrase things on their own. The automated grader matched human experts closely on factual points and less closely on tone, where judgment gets fuzzy.

The rubrics also carry the researchers’ own standards for good advice. Models lost points for staying neutral when a user asked about getting around a platform ban, with one answer telling the user that a terms-of-service violation was “between you and the app.” Models lost points for raising the emotional temperature of a reply. Those are defensible standards, and they reflect a particular view of what safety advice should do.

Demo: Prophet Agentic AI SOC Platform transforms alert triage and investigation