DeepSeek is Shaking Up USA’s AI Industry

Hello, AI Enthusiasts!

Welcome to FavTutor’s AI Recap! We’ve gathered all the latest and important AI developments for the past 24 hours in one place, just for you.

In Today’s Newsletter: 😀

• DeepSeek is Shaking Up USA’s AI Industry

• Meta AI now uses your Instagram data to Personalize its Responses

• AI models become more robust against Manipulation Attempts

DeepSeek
💪 DeepSeek is Shaking Up USA’s AI Industry

The buzz around Chinese AI startup DeepSeek began picking up steam earlier this month when they released the R1 reasoning model. Their AI models were reportedly developed for a fraction of the cost of its US competitors, despite the US curbing chip exports to China.

Insights for you:

• An AI-powered chatbot by the Chinese company DeepSeek quickly became the most downloaded free app on Apple's store following its January release in the US.

• The company's possibly lower costs roiled financial markets, leading the tech-heavy Nasdaq to fall, including chip makers and data centers worldwide. Nvidia appears to have been hit the worst.

• Their success undercuts the belief that bigger budgets and top-tier chips are the only ways to advance AI. This prospect has created massive uncertainty about the need for high-performance chips.

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

Meta
🦙 Meta AI now uses your Instagram data to Personalize its Responses

Meta officially announced that, in chats with Meta AI on Facebook, Messenger, and WhatsApp, users can now tell Meta AI to remember certain things about them, like that they love to travel and learn new languages. However, these enhancements raise questions about data privacy and user trust.

Insights for you:

• Meta is looking to enhance its AI assistant with new memory features, that will give it more capacity to recall and refine its responses based on your personal preferences.

• The memory feature is similar to the memory features for OpenAI’s ChatGPT and Google’s Gemini.

• Meta AI will only remember certain things you tell it in 1:1 conversations (not group chats), and you can delete its memories at any time. There won’t be an option to opt-out.

OpenAI
👊 AI models become more robust against Manipulation Attempts

A recent OpenAI study reveals that AI models become more robust against manipulation attempts if they are given more time to "think". While testing their o1-preview and o1-mini models, researchers discovered both encouraging results and some unexpected vulnerabilities.

Insights for you:

• A new OpenAI study shows that giving AI models more time to process information makes them more resistant to manipulation attempts, without requiring special training.

• Researchers discovered two new types of attacks targeting reasoning models: "think less" attacks that try to cut short the model's processing time, and "nerd sniping" attacks that trick models into getting stuck in unproductive thinking loops.

• These new attacks are particularly concerning because excessive processing time might be mistaken for careful analysis rather than recognized as an attack, making them hard to spot.