AI News for the week - 30th June
Lawyers have entered chat! Two of the frontier audio AI companies have just been served legal papers. Opensource hardware had another awesome week. The world’s first AI generated ad was released into the wild. OpenAI delayed its advance Voice mode and this upset people. OpenAI also announced new partnerships and a new desktop app for macOS. Character AI launched a new ‘Call’ feature. Stability AI has a new CEO and more money! Google opened up access to the 2 million token context window on Gemini 1.5 Pro for developers. Elevenlabs introduced their reader app and partnered with Time. And we put out a tutorial on how to use AI to make education a bit more fun. Let’s dive in!
Legal papers, that is. The music industry is fighting back against AI. The RIAA has filed a law-suit against Udio AI and Suno AI. They are claiming that these platforms have lifted music from various artists over various time periods to train their AI models. It’s probably true. This is the first big high profile law-suit against an AI company since the big AI boom. The outcome of this suit will set precedent for everything that happens in AI going forward. It could mean that all AI companies will need to obtain licenses to use any copyrighted works to train their models. This will extend much beyond music. Udio responded by saying that their platform doesn’t allow users to reference specific artists directly.
We’re big fans of Based Hardware. They’ve now launched a new wearable at $69.69. This one looks much better than the previous model and already has a marketplace for custom apps. Here’s Nik giving you the low-down:
Head on over to Basedhardware.com to make your purchase today!
ToysRUs partnered with OpenAI and used SORA their video generation model to create the world’s first AI generated ad:
This was met with some backlash from the creative community. The comments ranged from the fear of losing jobs to the genuinely poor quality of the ad. The ad itself is very uninspiring. It was clearly written for production in SORA and not specifically to tell the best story. The characters are not consistent. The storytelling is weak. The animations seem… off. We think this is the worst AI created ad ever. And that’s a good thing. It will only get better from here. Check out our POV on it here.
We love that OpenAI as a company ensures that they are always doing something worth noting each week. Despite that, people were upset when they delayed their voice model that was demoed in the spring update. It was delayed because it didn’t meet internal quality standards. Either that or the lawsuit against Suno and Udio spooked them. In either case, they can be cut some slack given that there are very few companies at their scale that ship updates and products as fast as they do. We are not complaining.
OpenAI also partnered with TIME to enhance responses and provide links to Time.com wherever relevant. They now have access to TIME’s 101 years of content. This is pretty dope and hopefully the right way of going about using copyrighted data.
OpenAI ALSO released a desktop app for macOS. This gives chatGPT the ability to give you information about whatever is visible on your screen. From e-mails, to screenshots to past conversations.
Known for being one of the most advanced human-like chatbots, Character AI announced that you could now also call and talk to your characters. Users can seamlessly switch between talking to their characters and texting them. This combined with their previous update on Character Voices makes for fun interesting AI companions. This all seems rather dark to us. Especially since human to human interaction is at an all time low. Have AI companions only makes matters worse.
Stability AI recently lost their CEO and founder along with a few lead engineers earlier in March. They’ve now come back with a new CEO, Prem Akkaraju, and have also secured ‘significant’ funding. Very required in the current AI arms race.
Stability AI – known for Stable Diffusion, has been faltering for a few months now with the community desirous of a better Stable Diffusion 3 model. With a new CEO, the hope is that they’ll get their ducks in a row and get back on the track for pushing the boundaries in generative AI.
This is good news for developers who are looking to work with AI that requires large amounts of context. Especially when it comes to research. The update also includes code execution which should improve the overall speed to outcome for developers. The clever part about this update is something called ‘code caching’. This is basically the ability for the AI to reuse code and context to help keep costs palatable (2 million tokens is a lot…). You can read more about it here.
Elevenlabs has been on a roll with their latest feature drops. This week they announced the Elevenlabs Reader App. You can now listen to any text (article, PDF, ePub) using a host of their AI voices.
The app is available on iOS and in the UK, USA and Canada with a waitlist for Android and other regions. You can sign-up here.
That’s it for the AI news update for this week. If you liked this and think that someone you know will find it useful, please share it with them. Also do visit us on LinkedIn. See you in the next one!
AI can be effectively used to teach English Grammar too. Here's a quick guide on…
If you're a teacher or a parent, you can use AI to help make learning…
One of the more hype weeks in AI in the last few months with new…
Apple Intelligence is pretty good. Open AI is in the news. Again! Luma Labs' Dream…
More hype in the AI Video space from China. Apple talks about AI the Apple…
Elon Musk and Yann LeCun troll each other on X. The first pilot episode of…