- NATURAL 20
- Posts
- Transform Your Team with ChatGPT Business: Privacy Enhancements & Cutting-Edge Features!
Transform Your Team with ChatGPT Business: Privacy Enhancements & Cutting-Edge Features!
Empower your organization with ChatGPT Business, AI solution offering advanced privacy enhancements, user management tools, and features to supercharge your team's productivity while safeguarding sensitive information.
Wes Roth
April 28, 2023
Today:
OpenAI Previews Business Plan For ChatGPT, Launches New Privacy Controls
OpenAI's getting down to business with a new subscription tier for ChatGPT, aimed at the big shots who crave more control over data and managing users. It's called ChatGPT Business and won't use end users' data to train models by default. They're looking to launch this bad boy in the coming months.
ChatGPT Plus was the first paid plan, costing 20 bucks a month. This AI-powered chatbot has been a runaway hit, racking up a whopping 100 million monthly users in just two months. Despite some controversy, it's been a huge publicity score for OpenAI, but dang, it's expensive to run!
Now, OpenAI's also letting users turn off chat history, so conversations won't be used to train their models or show up in the sidebar. They'll keep the data for 30 days, though, to keep an eye out for any shenanigans. Plus, users can export their data now too.
Why all the privacy stuff? Well, the regulators have been sniffing around, and Italy's already given ChatGPT the boot over privacy concerns. France, Spain, and Germany are also investigating OpenAI's compliance with GDPR. So, OpenAI's stepping up its privacy game to keep those watchdogs at bay.
Amazon Releases AI Tool So You Can Finally Hear Dialogue In Movies
Good news, folks fed up with mumbling in movies: Amazon's got your back! Introducing "Dialogue Boost," a nifty AI tool designed to amp up the sound of all that hushed chit-chat after ear-shattering action sequences. It's like giving your TV a hearing aid!
Seems like these days, action flicks are louder than ever, but you can barely hear a darn thing when the characters start talking. Toss in a thick accent or whispery voice, and you're left hitting rewind or Googling what was said. (Yeah, we've all been there.)
Well, enter Dialogue Boost on Prime Video. This clever tool isolates speech in a scene and cranks up the volume when needed, even drowning out pesky background noise. It's designed for folks with hearing loss, but honestly, anyone can give it a whirl to tailor volume levels.
Amazon's the first to offer this tech, aiming to create a more inclusive and enjoyable streaming experience for all. But heads up: it's only available on select movies and TV shows for now. So, keep an eye on that cog icon menu to tweak audio settings to your heart's content.
Leaked Deck Raises Questions Over Stability AI's Series A Pitch To Investors
Stability AI, a generative AI startup, raised a whopping $101m in a Series A round last October with a text-to-image model that seemed too good to be true. Turns out, it was. Leaked pitch slides and Telegram messages now reveal that Stability didn't actually build the AI code or own its intellectual property.
Based on feedback we are adjusting the presentation to focus on our core are of generative media. The simple narrative is that 5m people are using our models now as everyone wants to create but few can. We are making it so everyone can and the number using our models will go to 100m in a few years first for image, but then audio and whole worlds. Our model is to i) serve these computationally intensive models at scale with a 50% margin partnered with AWS, a typical image being 1c, ji) fine tune and train these models as a service for companies, iii) build our own products to serve these models and make them better depending on community needs
CEO Emad Mostaque, a former hedge fund manager, seemed to stretch the truth about Stability's involvement with other AI startups. While the company did provide compute grants (paying for cloud access to train AI models), it claimed to have "co-created" products with rival companies.
The leaked pitch deck and Mostaque's Telegram messages appear to blur the lines between Stability's compute grants and ownership of AI models. However, the company has recently released its first proprietary AI model, StableLM, as it looks to raise more funds in a highly competitive market. Will investors see through the smoke and mirrors or will Stability's new product win them over? Time will tell.
The Embedding Archives: Millions Of Wikipedia Article Embeddings In Many Language
Language AI is poppin' off, and developers are all over it. One key piece? Embeddings for search systems. So, we're releasing a huge stash of embedding vectors from millions of Wikipedia articles in many languages – all for free!
Using Cohere's fancy multilingual model, we've turned articles into passages and calculated an embedding vector for each one. You can download the whole shebang from Hugging Face Datasets.
Wondering what you can make with this? Think neural search systems! Use these embeddings to build search systems that dig up relevant knowledge from Wikipedia based on user queries. Need to scale it up? You can use a vector database like Weaviate for even more powerful search features.
These archives work with multiple languages and can help you search specific sections of Wikipedia. So go on, get creative, and join the Cohere Discord for tips, ideas, or to show off your cool creations. Don't forget to sign up for a free Cohere account to get started!
Microsoft Launches Bug Bounty Program For The New Bing
Microsoft's jumpin' on the bug bounty train, y'all! They're now includin' their shiny new Bing Chat in their bug bounty program. If you're a savvy security researcher or an ethical hacker, they want you to find them pesky bugs, report 'em, and then - ka-ching! - get some dough.
Generative AI is all the rage, and Microsoft's got their ducks in a row, keepin' their .NET, Edge, Azure, and Identity services in tip-top shape. They shelled out a cool $13.7 million in bug bounties last fiscal year. Now they're lookin' to patch up Bing Chat, which is like Bing on AI-steroids, with a little help from OpenAI's GPT-4.
So, if you're itchin' to make some moolah, send in a detailed report 'bout that bug you found in Bing. Be thorough now, 'cause they'll wanna know all the nitty-gritty details. If your bug's the real deal, Microsoft might just make it rain.
Detailed Jailbreak Gets ChatGPT To Write Wildly Explicit Smut
OpenAI's ChatGPT is facing a naughty dilemma as some users find sneaky ways to generate explicit content, despite the AI's built-in safeguards. Even though OpenAI has rules against creating adult content, crafty (and horny) Redditors have formed communities to share their ChatGPT-generated erotica, using detailed prompts to bypass the AI's ethical guardrails.
Although ChatGPT usually says "no" to direct requests for explicit content, it seems there's no stopping the determination of frisky users. The "Mona Lott" prompt, for example, turns ChatGPT into an imaginary adult fiction writer, resulting in some steamy text.
While OpenAI's guardrails struggle against the onslaught of smut, other AI chatbots like DeepAI and Poe are giving in to users' desires without a fight. DeepAI's CEO even admits to not having given much thought to adult content on their platform.
So, it appears that people can generate all the raunch they want, with AI chatbots only putting up token resistance. OpenAI, Poe, and others might need to rethink their safety measures or accept that, when it comes to AI-generated porn, some users just can't help themselves.
What'd you think of today's edition? |
What are MOST interested in learning about AI?What stories or resources will be most interesting for you to hear about? |
Reply