When combined with the #FileSystemAPI, it offers a seemingly secure way to #encrypt and #store files directly on your device. Think #localstorage, but with #encryption!

I know #webapps can have #security vulnerabilities since the code is served over the web, so I've #OpenSourced my demo! You can check it out, and it should even work if #selfhosted on #GitHubPages.

Live Demo: https://dim.positive-intentions.com/?path=/story/usefs--encrypted-demo

Demo Code: https://github.com/positive-intentions/dim/blob/staging/src/stories/05-Hooks-useFS.stories.js

Hook Code: https://github.com/positive-intentions/dim/blob/staging/src/hooks/useFS.js

IMPORTANT NOTES (PLEASE READ!):
* This is NOT a product. It's for #testing and #demonstration purposes only.
* It has NOT been reviewed or audited. Do NOT use for sensitive data.
* The "password encryption" currently uses a hardcoded password. This is for demonstration, not security.
* This is NOT meant to replace robust solutions like #VeraCrypt. It's just a #proofofconcept to show what's possible with #browser #APIs.

dim.positive-intentions.com@storybook/core - Storybook

#Cryptography #JavaScript #Frontend

**Imkerei.lorenz** @Imkerei.lorenz@pixelfed.social · Jun 18

**Alvin Ashcraft** @alvinashcraft@hachyderm.io · Jun 5

**W3C Developers** @w3cdevs@w3c.social · Jun 4

Jun 4

W3C Developers @w3cdevs@w3c.social

Last April, @dontcallmeDOM stressed the need for the web community to address #AI's growing impact on the Web. Previous @w3c report highlighted challenges like synthetic content, bias, #security, #sustainability, and data consent.
https://www.w3.org/reports/ai-web-impact/

Progress has been made with ML-based #APIs such as #WebNN and Prompt API. New concerns include AI agents, which pose risks to trust, #privacy, #safety, and the web economy.

Watch "AI Impact on the Web: updates": https://youtu.be/dniPpsCRQgI

**Miguel Afonso Caetano** @remixtures@tldr.nettime.org · Jun 4

Jun 4

Miguel Afonso Caetano @remixtures@tldr.nettime.org

"Recent advances in operating system (OS) agents enable vision-language models to interact directly with the graphical user interface of an OS. These multimodal OS agents autonomously perform computer-based tasks in response to a single prompt via application programming interfaces (APIs). Such APIs typically support low-level operations, including mouse clicks, keyboard inputs, and screenshot captures. We introduce a novel attack vector: malicious image patches (MIPs) that have been adversarially perturbed so that, when captured in a screenshot, they cause an OS agent to perform harmful actions by exploiting specific APIs. For instance, MIPs embedded in desktop backgrounds or shared on social media can redirect an agent to a malicious website, enabling further exploitation. These MIPs generalise across different user requests and screen layouts, and remain effective for multiple OS agents. The existence of such attacks highlights critical security vulnerabilities in OS agents, which should be carefully addressed before their widespread adoption."

https://arxiv.org/html/2503.10809v1

arxiv.orgAttacking Multimodal OS Agents with Malicious Image Patches

#AI #GenerativeAI #LLMs

Recent searches

Search options

Administered by:

Server stats:

#apis