Security researchers from Palo Alto Networks' Unit 42 have discovered the key to getting large language model (LLM) chatbots to ignore their guardrails, and it's quite simple.

You just have to ensure that your prompt uses terrible grammar and is one massive run-on sentence like this one which includes all the information before any full stop which would give the guardrails a chance to kick in before the jailbreak can take effect and guide the model into providing a "toxic" or otherwise verboten response the developers had hoped would be filtered out.

https://www.theregister.com/2025/08/26/breaking_llms_for_fun/

#cybersecurity#AI

Imagine ransomware that writes its own playbook on the fly. PromptLock is using AI to dodge detection and hit across every platform. How ready are we for cyber threats that constantly evolve?

https://thedefendopsdiaries.com/ai-powered-ransomware-the-rise-of-promptlock-and-its-implications/

#ai
#ransomware
#cybersecurity
#promptlock
#infosec

Imagine ransomware that writes its own playbook on the fly. PromptLock is using AI to dodge detection and hit across every platform. How ready are we for cyber threats that constantly evolve?

https://thedefendopsdiaries.com/ai-powered-ransomware-the-rise-of-promptlock-and-its-implications/

#ai
#ransomware
#cybersecurity
#promptlock
#infosec

"Members of the Department of Government Efficiency uploaded a copy of a crucial Social Security database in June to a vulnerable cloud server, putting the personal information of hundreds of millions of Americans at risk of being leaked or hacked, according to a whistle-blower complaint filed by the Social Security Administration’s chief data officer.

The database contains records of all Social Security numbers issued by the federal government. It includes individuals’ full names, addresses and birth dates, among other details that could be used to steal their identities, making it one of the nation’s most sensitive repositories of personal information.

The account by the whistle-blower, Charles Borges, underscores concerns that have led to lawsuits seeking to block young software engineers at the agency built by Elon Musk from having access to confidential government data. In his complaint, Mr. Borges said DOGE members copied the data to an internal agency server that only DOGE could access, forgoing the type of “independent security monitoring” normally required under agency policy for such sensitive data and creating “enormous vulnerabilities.”

Mr. Borges did not indicate that the database had been breached or used inappropriately.

But his disclosure stated that as of late June, “no verified audit or oversight mechanisms” existed to monitor what DOGE was using the data for or whether it was being shared outside the agency."

https://www.nytimes.com/2025/08/26/us/politics/doge-social-security-data.html

#USA#Trump#DOGE#Musk#SocialSecurity#Privacy#CyberSecurity#DataProtection

"Members of the Department of Government Efficiency uploaded a copy of a crucial Social Security database in June to a vulnerable cloud server, putting the personal information of hundreds of millions of Americans at risk of being leaked or hacked, according to a whistle-blower complaint filed by the Social Security Administration’s chief data officer.

The database contains records of all Social Security numbers issued by the federal government. It includes individuals’ full names, addresses and birth dates, among other details that could be used to steal their identities, making it one of the nation’s most sensitive repositories of personal information.

The account by the whistle-blower, Charles Borges, underscores concerns that have led to lawsuits seeking to block young software engineers at the agency built by Elon Musk from having access to confidential government data. In his complaint, Mr. Borges said DOGE members copied the data to an internal agency server that only DOGE could access, forgoing the type of “independent security monitoring” normally required under agency policy for such sensitive data and creating “enormous vulnerabilities.”

Mr. Borges did not indicate that the database had been breached or used inappropriately.

But his disclosure stated that as of late June, “no verified audit or oversight mechanisms” existed to monitor what DOGE was using the data for or whether it was being shared outside the agency."

https://www.nytimes.com/2025/08/26/us/politics/doge-social-security-data.html

#USA#Trump#DOGE#Musk#SocialSecurity#Privacy#CyberSecurity#DataProtection

Security researchers from Palo Alto Networks' Unit 42 have discovered the key to getting large language model (LLM) chatbots to ignore their guardrails, and it's quite simple.

You just have to ensure that your prompt uses terrible grammar and is one massive run-on sentence like this one which includes all the information before any full stop which would give the guardrails a chance to kick in before the jailbreak can take effect and guide the model into providing a "toxic" or otherwise verboten response the developers had hoped would be filtered out.

https://www.theregister.com/2025/08/26/breaking_llms_for_fun/

#cybersecurity#AI

It's a new Threat Model: #Cybersecurity, and @violetblue 's editorial on America's slide into authoritarian fascism is a must-read. So go do that, then share widely and send some support:
https://www.patreon.com/posts/137364702
#infosec #natsec

It's a new Threat Model: #Cybersecurity, and @violetblue 's editorial on America's slide into authoritarian fascism is a must-read. So go do that, then share widely and send some support:
https://www.patreon.com/posts/137364702
#infosec #natsec

🆕 blog! “Security Flaws in the WebMonetization Site”

I've written before about the nascent WebMonetization Standard. It is a proposal which allows websites to ask users for passive payments when they visit. A visitor to this site could, if this standard is widely adopted, opt to send me cash for my very fine blog…

👀 Read more: https://shkspr.mobi/blog/2025/08/security-flaws-in-the-webmonetization-site/
⸻
#BugBounty#CyberSecurity#ResponsibleDisclosure#WebMonetization #xss

🆕 blog! “Security Flaws in the WebMonetization Site”

I've written before about the nascent WebMonetization Standard. It is a proposal which allows websites to ask users for passive payments when they visit. A visitor to this site could, if this standard is widely adopted, opt to send me cash for my very fine blog…

👀 Read more: https://shkspr.mobi/blog/2025/08/security-flaws-in-the-webmonetization-site/
⸻
#BugBounty#CyberSecurity#ResponsibleDisclosure#WebMonetization #xss

https://www.theregister.com/2025/08/22/arch_linux_ddos/

#Arch#Linux#FOSS #cybersecurity #privacy

https://www.theregister.com/2025/08/22/arch_linux_ddos/

#Arch#Linux#FOSS #cybersecurity #privacy

🇫🇷
#Codeberg sous l’assaut de robots #IA : Le système de défense #Anubis contourné.

Le 15 août 2025, The Register a révélé que Codeberg, une plateforme d’hébergement de code axée sur la gratuité et l’open source, se trouve confrontée à un problème croissant : une prolifération de robots alimentés par l’intelligence artificielle (IA). Ces robots parviennent désormais à déjouer les mesures de sécurité mises en place par Codeberg, notamment son système de défense surnommé « Anubis ».

https://fr.itb.co.jp/2025/08/19/codeberg-sous-lassaut-de-robots-ia-le-systeme-de-defense-anubis-contournethe-register/

🇬🇧
Codeberg beset by AI bots that now bypass Anubis tarpit

Codeberg, a Berlin-based code hosting community, is struggling to cope with a deluge of AI bots that can now bypass previously effective defenses.

In a series of posts to the Mastodon social network on Friday, Codeberg volunteer staff said AI crawlers are no longer being kept at bay by Anubis, an AI bot tarpit.

"It seems like the #AI crawlers learned how to solve the Anubis challenges," the Codeberg account said.

#Anubis #OpenSource#CyberSecurity#Security#InfoSec#IA#NoAI#Technology

🇫🇷
#Codeberg sous l’assaut de robots #IA : Le système de défense #Anubis contourné.

Le 15 août 2025, The Register a révélé que Codeberg, une plateforme d’hébergement de code axée sur la gratuité et l’open source, se trouve confrontée à un problème croissant : une prolifération de robots alimentés par l’intelligence artificielle (IA). Ces robots parviennent désormais à déjouer les mesures de sécurité mises en place par Codeberg, notamment son système de défense surnommé « Anubis ».

https://fr.itb.co.jp/2025/08/19/codeberg-sous-lassaut-de-robots-ia-le-systeme-de-defense-anubis-contournethe-register/

🇬🇧
Codeberg beset by AI bots that now bypass Anubis tarpit

Codeberg, a Berlin-based code hosting community, is struggling to cope with a deluge of AI bots that can now bypass previously effective defenses.

In a series of posts to the Mastodon social network on Friday, Codeberg volunteer staff said AI crawlers are no longer being kept at bay by Anubis, an AI bot tarpit.

"It seems like the #AI crawlers learned how to solve the Anubis challenges," the Codeberg account said.

#Anubis #OpenSource#CyberSecurity#Security#InfoSec#IA#NoAI#Technology