SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Ghost Font claims to be readable by humans but invisible to AI systems

0
·2 views

A new typeface called Ghost Font has been developed with the claim that humans can read it while AI systems cannot. The font appears on the MixFont platform and has attracted attention in developer and tech circles. The concept targets growing concerns around AI-based text recognition and automated data scraping. While details on the underlying technique remain limited, the project has sparked discussion about the feasibility and practical applications of AI-resistant typography.

Read the full story at Hacker News

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Developer builds free offline Windows image converter after finding existing tools too complex

A solo developer under Akhouri Systems has released APIC, a free Windows desktop tool designed to simplify bulk image conversion. The tool was created after its author found common alternatives — such as online converters, GIMP, and Photoshop — too cumbersome, requiring multiple steps or large installations. APIC supports conversion between formats including PNG, JPG, WEBP, BMP, TIFF, and ICO, and also offers editing, compression, and batch processing features. The application runs entirely offline as a single .exe file, meaning files never leave the user's machine. APIC is available for free download via GitHub under the Akhouri Systems project.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Slack Bot That Autonomously Writes Code and Opens GitHub Pull Requests

A software engineer created Slack2PR, an AI-powered bot that integrates directly into Slack to automate the process of turning ideas or bug reports into GitHub pull requests. Users simply mention the bot in a Slack channel, describe their request, and the agent handles requirements gathering, planning, implementation, testing, and PR creation. The system classifies requests into three categories — features, bugs, and questions — each triggering a distinct workflow with appropriate approval steps. Built on Hexabot, it uses Google Gemini for AI tasks, OpenCode as the coding engine, and TanStack AI Sandboxes to run each job in an isolated Docker environment. The project was developed as part of the DEV Weekend Challenge to demonstrate how AI agents can be embedded into everyday development workflows rather than used as standalone chat tools.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to secure Terraform state storage without a VPN using zero-trust networking

Terraform state files are among the most sensitive cloud artifacts, often storing secrets in plain text, yet they are commonly hosted on publicly accessible storage endpoints protected only by an access key. A two-phase bootstrap approach solves the chicken-and-egg problem of securing this storage: the first phase creates the state backend infrastructure using local state, while the second phase migrates to remote state and locks down public network access. Rather than deploying a costly VPN gateway or bastion host to allow CI pipelines and operators to reach the now-private storage, a lightweight zero-trust connector runs as a container inside the virtual network. This connector joins an identity-aware mesh, granting authenticated access to the private endpoint only for the duration needed, with no certificates to rotate or permanent infrastructure to maintain. The key pitfall to avoid is disabling public access before the private network route exists, which would cut off Terraform mid-apply before it can write state.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

FFmpeg Micro Offers Cloud API Alternative to fluent-ffmpeg for Node.js Serverless Apps

Node.js developers using fluent-ffmpeg face significant challenges when deploying to serverless platforms like AWS Lambda, Vercel, or Google Cloud Functions, which do not include FFmpeg binaries by default. Bundling a static FFmpeg build adds 70–100MB to deployment size and introduces architecture-specific fragility, while a single 1080p transcode can demand 2–4GB of RAM. Cold starts are further delayed by 2–5 seconds due to binary loading on top of standard Node.js initialization. FFmpeg Micro is a cloud API designed to address these issues by running FFmpeg commands on dedicated remote infrastructure via a simple HTTP call. The service eliminates the need for binary installation, local memory management, or Lambda Layers, and is accessible from any runtime that supports the fetch API.

0 comments Read more at DEV Community