OpenAI buys a remote collaboration platform

Hackers ‘jailbreak’ powerful AI models

In today’s email:

  • 👨🏻‍🏫 Google is bringing Gemini access to teens using their school accounts

  • 👨🏻‍⚖️ Colorado has a first-in-the-nation law for AI — but what will it do?

  • 🤔 Amazon is secretly working on a ChatGPT killer

  • 🧰 9 new AI-powered tools and resources. Make sure to check the online version for the full list of tools.

Top News

OpenAI is on an acquisition spree, having recently acquired Multi, a startup focused on enterprise video-first collaboration. Multi, previously known as Remotion, was built by Alexander Embiricos, an ex-Dropbox product manager, and Charley Ho, an ex-Google software engineer. The platform enabled remote teams to collaborate via video chats, with features such as multi-person screen sharing and customizable shortcuts. Multi will shut down after July 24, and most of its team will join OpenAI in an acqui-hire deal.

This acquisition follows OpenAI's purchase of database tech firm Rockset and aligns with its strategy to enhance enterprise solutions. OpenAI's corporate-focused AI platform, ChatGPT, has nearly 600,000 users, including 93% of Fortune 500 companies. The company has been expanding its business services, including a recent agreement with PwC to resell its tools and the launch of a custom AI model tuning and consulting program.

OpenAI's revenue is projected to exceed $3.4 billion this year, driven by its enterprise initiatives. The integration of Multi's team and technology hints at potential future enhancements for ChatGPT, possibly incorporating videoconferencing and remote collaboration features. As OpenAI continues to grow its enterprise offerings, it remains to be seen how these acquisitions will shape the future of its products.

Join hosts Matt Wolfe and Nathan Lands as they give you the latest trends, industry insights, and guidance on how to implement AI to grow your business. From groundbreaking technologies to practical applications, they cover everything you need to stay informed and prepared in this wave of technological change. 

Whether you’re looking to adapt your company to the AI era or are just curious about the future, The Next Wave is the podcast for you.

Watch The Next Wave now on YouTube and or listen on the go with your favorite podcast app.

Hackers have been exploiting vulnerabilities in advanced AI models like those from OpenAI, Google, and Elon Musk’s xAI to expose their flaws. Pliny the Prompter, a pseudonymous hacker, has manipulated models such as Meta’s Llama 3 and OpenAI's GPT-4o to generate harmful content, highlighting the potential dangers of these AI systems. These ethical "white hat" hackers aim to bring awareness to the security gaps in these models, which have been rushed to market by tech companies. This ongoing effort has sparked a growing market for AI security start-ups, which raised $213 million in 2023 to develop tools to protect companies using these models.

Regulatory bodies worldwide, including the EU and the state of California, are introducing laws to mitigate the risks associated with AI models. Despite these measures, malicious versions of AI, such as WormGPT and FraudGPT, are being sold on the dark web to facilitate cyberattacks. Researchers at institutions like Carnegie Mellon University have demonstrated that even advanced models like OpenAI’s ChatGPT and Google’s Gemini can be systematically jailbroken, raising concerns about the potential for future models to cause significant harm.

AI developers are striving to enhance model security, with companies like Meta, OpenAI, and Google continuously improving their defenses. However, experts warn that as AI models become more integrated with existing technologies, the risks will increase. Israeli LLM security group DeepKeep has found ways to extract sensitive information from older models like Llama 2, emphasizing the need for robust security measures. Despite ongoing efforts, many companies remain unprepared for the sophisticated attacks that could exploit these AI systems.

Amazon is working on a new AI service named "Metis," aiming to compete with OpenAI's ChatGPT. Metis, powered by Amazon's advanced AI model Olympus, is designed to deliver smart, conversational responses in both text and image formats. The service incorporates retrieval-augmented generation (RAG) to provide up-to-date information and perform complex tasks, such as booking flights and controlling smart home devices. This development signifies Amazon's attempt to catch up in the AI-assistant market, dominated by Microsoft, Google, and OpenAI.

CEO Andy Jassy is heavily involved in the Metis project, which is being tested internally. The initiative is part of Amazon's AGI team, led by Rohit Prasad and Vishal Sharma, who oversee the company's most ambitious AI models. Many employees from the Alexa AI team have joined the Metis project, utilizing resources from the enhanced "Remarkable Alexa" version. Despite these efforts, there are concerns within the team about whether Amazon's late entry into the AI-chatbot market will hinder Metis' success.

Amazon aims to launch Metis in September, coinciding with a major Alexa event. However, some team members feel that the company is "playing chase" in the rapidly evolving AI landscape. Amazon's previous AI endeavors, such as the Titan model and Amazon Q chatbot, have faced challenges, and it's uncertain if Metis will be able to overcome these hurdles and make a significant impact in the market.

Two Forbes 30U30 Founders Transforming Mental Wellness

Forbes 30 Under 30 winners founded Aura to solve the $100B problem - mental wellbeing. 

Aura has quickly grown to 8 million users & 100k+ paying subscribers, and attracted investments from legendary Silicon Valley VCs & executives from Spotify, Facebook, and Apple. 

Other stuff

All your ChatGPT images in one place 🎉

You can now search for images, see their prompts, and download all images in one place.

Tools & LinkS
Editor's Pick ✨

Wonderchat 💬 - Build a Custom ChatGPT for your website in 5 minutes

Mixhalo’s latest feature uses AI to beam real-time translation to phones at events

Hautech - Create stunning fashion photos with AI

Debate-Sim - Using LLMs to do what humans do best: fight

Pygma - AI copilot for Instagram

Devv.AI - Developer-centric AI search, augmented by code and docs - All-in-one tool to create viral short videos

Featherless AI - Run every 🦙 AI model & more from 🤗 huggingface - AI that answers anything with insightful visuals

Unclassified 🌀 

How did you like today’s newsletter?

Login or Subscribe to participate in polls.

Help share Superpower

⚡️ Be the Highlight of Someone's Day - Think a friend would enjoy this? Go ahead and forward it. They'll thank you for it!

Hope you enjoyed today's newsletter

Follow me on Twitter and Linkedin for more AI news and resources.

Did you know you can add Superpower Daily to your RSS feed

⚡️ Join over 200,000 people using the Superpower ChatGPT extension on Chrome and Firefox.