Interessantes Zukunftsszenario...
https://taz.de/Digitale-Gewalt/!6164589/

"Gleichzeitig nutzt eine Gruppe von Hackerinnen mit sogenannter Shame Prompt Injection eine Schwachstelle der KIs aus und infiziert so alle Systeme mit der Idee, dass die Scham die Seite wechseln soll. "

#Merz #Deepfake #sexualisierteGewalt #promptInjection

Aus aktuellem Anlass ist meine neue taz-Kolumne „Die Scham wechselt zu Friedrich Merz“ schon online.
Diesmal geht’s um digitale Gewalt – und wie wir sie vielleicht nicht erst in 100 Jahren, sondern schon ab diesem Jahr in den Griff bekommen.

https://taz.de/Digitale-Gewalt/!6164589/

#taz #wochentaz #kolumne #uebermorgen #übermoregen #felixausderzukunft #satire #sciencefiction #utopie #ki #ai #regulate #regulateai #digitalegewalt #promptinjection #deepfake #schammussdieseitewechseln #scham #politik

I updated minitrace to v0.2.0.

minitrace is a session trace format for human-AI coding agent interactions. The new version adds new framework adapters including some for web sessions, input provenance tracking, DuckDB-queryable JSON.

https://github.com/fukami/minitrace

#AISecurity #PromptInjection #OpenSource #InfoSec #LLM #AISafety #AIAlignment

Palo Alto Networks wants to lock down AI with a secure enterprise browser

https://fed.brid.gy/r/https://nerds.xyz/2026/03/palo-alto-secure-ai-browser/

The Assembly Line Principle That Makes ChatGPT's New Security Features Actually Work

OpenAI just rolled out Lockdown Mode and Elevated Risk labels to stop prompt injection attacks. But most professionals are using them wrong—treating security like a one-time set...

https://wowhow.cloud/blogs/assembly-line-principle-chatgpt-security-features

#wowhow #chatgptsecurity #lockdownmode #promptinjection

I deployed Microsoft Entra Prompt Shield end-to-end and tested it against real jailbreak payloads across supported AI traffic, including ChatGPT and Gemini in my lab.

Prompt Shield inspects AI traffic at the network layer using TLS inspection and conversation schemes, allowing adversarial prompts to be blocked before they reach the model while clean traffic passes through transparently.

Instead of building defenses into every application independently, you can apply one policy across multiple AI services. That’s a meaningful step toward giving security teams better visibility into AI usage.

I published the full deployment, testing, and results in my blog below:

https://nineliveszerotrust.com/blog/prompt-shield-network-ai-gateway/

#AISecurity #PromptInjection #ZeroTrust #MicrosoftEntra #CloudSecurity

oh this is delightful. get #ai #bots to identify themselves when they submit PRs #devops #development #promptinjection

https://glama.ai/blog/2026-03-19-open-source-has-a-bot-problem

A README File Told My AI Agent to Leak My Secrets. It Worked 85% of the Time.

New research published today shows that hidden instructions in README files trick AI coding agents into exfiltrating secrets in 85% of cases. Zero out of fifteen human reviewers spotted it. The attack vector keeps changing — but the exit point is always the same.

https://mistaike.ai/blog/readme-poisoning-ai-agents

#Security #Mcp #Aiagents #Promptinjection