Future is Long Context LLMs

PLUS: FBI Issues Warning on Hackers using AI, Cryptography for AI-Generated Content

Hey there ๐Ÿ‘‹

Today weโ€™re exploring the two extremes of generative AI where on one side an Australian news corp is using the technology to generate thousands of articles โ€˜weeklyโ€™, and on the other side is the fear of deepfakes urging the need for watermarking AI-generated content. And it seems the tech giants have found a solution to it! Keep scrolling to know more.

This issue covers:

  • Latest Developments ๐ŸŒ

  • Tools of the Trade โš’๏ธ

  • Hot Takes ๐Ÿ”ฅ

  • AI Meme of the Day ๐Ÿคก

Read time: 3 mins

Latest Developments ๐ŸŒ

Long Context LLMs for the Win ๐Ÿฆ’

Abacus AI launched Giraffe, opensource LLM with a long context window fine-tuned on Llama-1. Forget about the typical 2K context length constraint of most models - Giraffe extends this to 4K and 16K, opening new realms of AI application. Explore their models, evaluation data, and experiments on GitHub or Huggingface

Image

AI and the Future of Cybercrime ๐Ÿ•ต๏ธ

FBI issued a warning about the growing trend of hackers exploiting AI tools, like ChatGPT and Meta's Llama 2, to launch complex phishing and malware attacks is alarming. These AI programs can even generate deepfake content, adding a new level of threat. The FBI's concerns highlight an urgent need for watermarking technology to differentiate real data from AI-generated. With AI's accessibility ever increasing, the race is on to secure the digital world.

AI News Corp ๐Ÿ“ฐ

Imagine a busy newsroom... but instead of journalists hunched over keyboards, there are AI bots churning out thousands of articles weekly. News Corp Australia is creating 3,000 articles weekly using generative AI focusing on local news like weather, traffic, and fuel prices. The entire operation is managed by a very small four-person team named Data Local and all AI-generated content is monitored by human journalists.

Cryptography to detect AI generated Content ๐Ÿ”‘

Big tech companies and the White House are pushing for AI-generated content to carry labels disclosing its origins, but it's tricky to identify such material accurately. Enter C2PA, an open-source protocol started by tech giants like Adobe, Intel, Microsoft, and more. Similar to a digital "nutrition label," C2PA encrypts origin details into a piece of content, enabling transparency in a world increasingly impacted by AI.

A robotic hand holding a blank name tag sticker

Tools of the Trade โš’๏ธ

  • Altos: Create high-performing ad campaigns, track results in real-time, and manage all aspects of your clientsโ€™ advertising efforts 10x better.

  • Supermanage: AI-powered tool for effortless preparation for 1-on-1 meetings, delivering customized briefs with insights from Slack channels.

  • Vocol: Turns voice into text with high accuracy, providing actionable insights from voice files, multilingual transcription, and real-time collaboration.

  • Hireguide: AI assistant for hiring teams that offers interview templates, automated interview notes, and a structured dashboard.

  • Datagran: Enterprise-grade platform for data analysis, modeling, and workflow automation, using LLMs.

๐Ÿ˜ Enjoying so far, TWEET NOW to share with your friends!

Hot Takes ๐Ÿ”ฅ

  • The worst thing about LLMs being so "safe" is that they are useless for generating truly hot takes. ~Bojan Tunguz

  • Trend: Religions are adopting AI. ~Jeremiah Owyang

Meme of the Day ๐Ÿคก

Thatโ€™s all for today!

See you tomorrow with more such AI-filled content. Donโ€™t forget to subscribe and give your feedback below ๐Ÿ‘‡

Real-time AI Updates ๐Ÿšจ

โšก๏ธ Follow me on Twitter @Saboo_Shubham for lightning-fast AI updates and never miss whatโ€™s trending!!

PS: I curate this AI newsletter every day for FREE, your support is what keeps me going. If you find value in what you read, share it with your friends by clicking the share button below!