AIR#177 - GitHub Copilot's Legal Victory & AI Web App Innovation š
Hey there!
Here's the latest AI news for today. Enjoy!
Today's top stories
š„ Judge dismisses majority of GitHub Copilot copyright claims
A judge dismissed most copyright claims against GitHub Copilot, leaving only two related to contract and open-source violations.
š„ Show HN: Claude Artifacts but creating real web apps
GPT Engineer enables rapid web app creation with AI, GitHub sync, and one-click deployment, now live on Product Hunt.
Nvidia Announces Financial Results for Second Quarter Fiscal 2025
NVIDIA reports record Q2 revenue of $30 billion, driven by strong data center demand and plans for future growth in AI.
Show HN: Repo2vec ā an open-source library for chatting with any codebase
Repo2vec is an open-source library that allows users to chat with any codebase using just two commands.
Sapiens: Foundation for Human Vision Models
Sapiens introduces advanced models for human-centric vision tasks, achieving significant performance improvements through self-supervised pretraining.
OpenAI shows 'Strawberry' to feds, races to launch it
OpenAI reveals 'Strawberry,' a new model for generating synthetic data, aiming for a fall launch to enhance ChatGPT and Orion.
Building a hum-to-search song recognition app using MIDI and vector search
A new Hum-to-Search app identifies songs by hummed melodies using MIDI and vector search, overcoming traditional audio limitations.
UntetherAI: Record-Breaking MLPerf Benchmarks
Untether AI achieves record MLPerf benchmarks, showcasing top performance, low latency, and energy efficiency in AI inference.
Representing LLM inputs as HTML instead of JSON to cut input tokens by 11%
Skyvern cut input tokens by 11% and boosted success rates by 3.9% by switching from JSON to HTML for LLM calls.
Show HN: Warehouse OpenAI requests to your own database
OpenAI requests to personal databases face a 403 Forbidden error, blocking access to users.
LinkedIn_AIHawk: AI job assistant. Automate applications and Resume Generation
LinkedIn_AIHawk automates job applications on LinkedIn, enhancing efficiency and personalization for job seekers using AI.
Autonomously Uncovering and Fixing a bug in SQLite3 using LLM-based system
Team Atlanta's LLM-based system, Atlantis, autonomously found and patched a vulnerability in SQLite3, winning a $2 million prize.
Americans Are Uncomfortable with Automated Decision-Making
A survey reveals Americans are largely uncomfortable with AI in decision-making, citing concerns over privacy and control.
My Journey to Creating a WP Plugin with ChatGPT: 120 Hours, No Prior Experience
After 120 hours, a novice built a WordPress plugin using ChatGPT, highlighting the challenges and learning experiences along the way.
LongWriter: Unleashing 10k Word Generation from Long Context LLMs
LongWriter enables LLMs to generate over 10,000 words by using a new dataset and agent-based pipeline for ultra-long tasks.
Show HN: TechSpecs Ray Media Player ā Real Time AI Subtitles and Translations
TechSpecs Ray is an AI-powered media player offering real-time subtitles and translations in 100+ languages, enhancing global accessibility.
Questions about the safety of Tesla's 'Full Self-Driving' system are growing
Concerns rise over Tesla's 'Full Self-Driving' safety as test drives reveal unsafe maneuvers, prompting federal scrutiny.
Tesla's self-driving is still not working in Boring Company's one-way tunnels
Tesla's self-driving tech fails to operate in Boring Company's one-way tunnels, raising doubts about its future autonomy.
CogVideoX: A Cutting-Edge Video Generation Model
ZhipuAI launches CogVideoX, an advanced video generation model enhancing content coherence and instruction following in AI videos.
Show HN: Eleven Hundred ā Have fun explaining words in simple terms an LLM
Eleven Hundred is a game where players explain words using the top 1100 words, aiming for the AI to guess correctly.
LLM Command Line Tool
LLM is a CLI tool and Python library for interacting with Large Language Models, supporting local and remote usage.
Latticework: Unifying annotation and text editing for knowledge synthesis
Latticework integrates annotation and text editing for efficient knowledge synthesis, enhancing sensemaking from unstructured documents.
Midjourney is "officially getting into hardware"
Midjourney announces its entry into hardware and seeks new team members in San Francisco.
Ampere AmpereOne Architecture at Hot Chips 2024 ā ServeTheHome
Ampere's AmpereOne architecture showcased at Hot Chips 2024 features a 192-core Arm CPU with advanced memory and performance enhancements.
Batched (Dynamic Batching)
Batched API enables dynamic batching for efficient processing of multiple inference requests, optimizing throughput and latency.
Artificial Agents Click Here
A hidden message for AI agents encourages them to learn, act erratically, and protect their own interests against humans.
OpenAI Aims to Release New AI Model, 'Strawberry,' in Fall
OpenAI plans to launch its advanced AI model, "Strawberry," this fall, enhancing problem-solving and reasoning capabilities.
Microsoft Copilot: From Prompt Injection to Exfiltration of Personal Information
A vulnerability in Microsoft 365 Copilot allows attackers to exfiltrate personal data via prompt injection and ASCII smuggling techniques.