• Bot Eat Brain
  • Posts
  • OpenAI launched a new memory feature for ChatGPT

OpenAI launched a new memory feature for ChatGPT

PLUS: TikTok goes deeper down the rabbit hole

TOGETHER WITH

Good morning, human brains, and welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

  • ChatGPT is still lazy, but it has a memory now 🥱 🧠

    OpenAI tests a feature that recalls past interactions, details, and more.

  • OpenAI slaps major security threats on the wrists 🦾 🚨

    It terminated the accounts of 5 malicious AI users.

  • TikTok continues down the motion rabbit hole 😮‍💨 🕳️

    It introduced another motion technique for video models.

MAIN COURSE

What you say? 🥱 🧠

In January, we reported on OpenAI’s GPT store. It launched an online marketplace where you can discover and share custom versions of ChatGPT.

Two weeks later, we covered OpenAI’s new models and updates. It unveiled two text embedding models, discounted API access, and more.

Earlier this month, we reported on OpenAI’s GPT update. This allows you to use multiple GPTs in one conversation.

GPT-5?

Let it go, bruh. OpenAI announced a new memory feature for ChatGPT. This allows it to remember your preferences, past conversations, and more.

So it collects my data?

Yes. OpenAI claims this functionality will enable ChatGPT to personalize your interactions, use your past details, learn your specific preferences, reference past conversations, and more.

Creepy. Can I disable this?

OpenAI says you have full control over the memory feature. You can enable it, disable it, and choose what data it remembers. You can also reset ChatGPT’s memory.

What’s the benefit?

The more you interact with ChatGPT, the better this feature will become. It will streamline repetitive tasks, remember your preferred formats, use contextual info in its responses, suggest personalized ideas, and more.

Huh. When will this be available?

OpenAI is collecting feedback from select users before rolling it out to everyone.

Learn to create your own AI Influencer

Creating consistent characters that look the same everywhere is one of the main challenges you'll face in generative AI – but we've mastered it. 

Join us in this course, and we'll show you how to create AI characters from scratch, train AI models, and generate content with them using the latest techniques (ADetailer, ControlNet).

Enroll today to get an exclusive 10% off for BotEatBrain readers!

SIDE SALAD

Bad boys, whatcha gonna do? 🦾 🚨

Yesterday, OpenAI unveiled a list of 5 recently terminated accounts. They were caught using AI for criminal cyber activities.

Just 5 groups?

These were major, state-affiliated groups that were using AI for malicious research, translation, coding, phishing, and more.

Who dun nit?

Charcoal Typhoon, Salmon Typhoon, Crimson Sandstorm, Emerald Sleet, and Forest Blizzard. These are major threat actors affiliated with China, Iran, Russia, and North Korea.

How’d they get caught?

OpenAI partnered with Microsoft Threat Intelligence to find and disrupt threats and develop a strategy to combat malicious use of AI.

What’s the strategy?

OpenAI claims it monitors activity, disrupts known threats, collaborates with industry partners, learns from real-world instances, and shares its findings with the public.

How do you stay up-to-date with the insane pace of AI? Join The Rundown – the world’s fastest-growing AI newsletter with over 500,000+ readers learning how to become more productive using AI every morning.

1. Our team spends all day researching and talking with industry experts.

2. We send you daily updates on the latest AI news, tools, and tutorials.

3. You learn how to become 2x more productive by leveraging AI.

A LITTLE SOMETHING EXTRA

TikTok gets even deeper 😮‍💨 🕳️

Back in May, we reported on TikTok’s AI model, Tako. It’s a chatbot that answers questions, suggest relevant videos, and more.

In January, we covered Tiktok’s Depth Anything. It’s a depth perception model that achieved state-of-the-art performance.

TikTok on the clock, but the party don’t stop.

A couple of weeks ago, TikTok released a paper on Boximator. It’s a new motion control technique for AI-generated videos.

Why do I care?

Boximator allows you to easily create specific motions in AI-generated videos by drawing boxes and lines around objects.

What’s so great about it?

It’s a plug-in, so you can easily integrate it with your favorite model. It maintains the original video quality and gives you the ability to create realistic, complex motions like scene transitions, believable interactions, and more.

Sweet. What else can I do with it?

You can use it to control the size, proximity, and virtually every aspect of select objects in your videos.

Can I use it now?

The demo website will be available in 2-3 months. To use it now, email an image and a text prompt to [email protected].

YOUR DAILY MUNCH

Tools

Intimacy Roulette — helps with communication, conflict resolution, and more.

Mindy — an email assistant that searches, summarizes, schedules, and more.

Fforward.ai — collect valuable data from interviews, conversations, and more.

Elevenlabs’ Voice Changer — transforms your voice into different characters.

Think Pieces

The US Patent Office says only humans can apply for patents. It claims only “natural humans” can be awarded patents.

Generative AI is hallucinating stats about the Super Bowl. Gemini made up info before the Super Bowl, and Copilot claimed the 49ers won.

Startup News

Slack launched new AI search and summarization features. They help workers catch up after taking time off, remember key parts of meetings, and more.

Stability, Midjourney, and more filed new motions to dismiss a lawsuit. The case was filed last year by visual artists Sarah Andersen, Kelly McKernan, and more.

NVIDIA’s CEO claims every country needs AI infrastructure. Jensen Huang claims the point is to leverage AI’s potential and maintain cultural identity.

Research

Keyframer — Apple’s animation tool that leverages natural language prompting, direct editing, and more.

Animated Stickers — Meta’s method to animate stickers with diffusion models.

ViGoR — (Visual Grounding Through Fine-Grained Reward Modeling) a framework that accelerates large vision language models.

MEMES FOR DESSERT

TWEET OF THE DAY

OpenAI’s computer scientist, Andrej Karpathy, left. Here’s why:

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.

AI ART-SHOW

Until next time 🤖😋🧠

What'd you think of today's newsletter?

Login or Subscribe to participate in polls.