📰 Welcome back!

Apple’s ‘Awe Dropping’ September event is just days away where they’ll be unveiling new AirPods or something else for you to cram into another orifice.

It’s their annual event where it’s out with the old and in with the new. But while its pats on the backs for being “courageous” and removing the volume buttons on the side of a phone, they won’t be so vocal about the defections at their own AI department.

That’s right, the AI employee exodus at Apple has continued with four more workers leaving in order to join rival firms. But are Apple worried? Probably not. Well, not yet anyway…

🚀 What we’re covering today…

• 🍎 The Apple AI employee exodus continues

• 🚗 Tesla unveiled Part IV of its ‘Master Plan’

• ⚖️ Google told to stop being so god damn anticompetitive

• 🛍️ Amazon launches Lens Live to help you with shopping

• 🖕 Cracks start to show in Meta and Scale AI relationship

• 🇨🇳 China makes new mandate on labelling all AI content

🔴 Quick Note: We like to cover loads of AI news in our newsletter, so for a better reading experience, we suggest opening this in your browser for the full experience! 

Head to the ‘READ ONLINE’ tab at the top of this email.

👁️ 👁️ What you might have missed

• “It’s happened again! It’s happened again! Apple AI department, it’s happened again!” Not the catchiest of football chants and doesn’t quite roll off the tough but the point still stands: Apple have lost more employees to rivals firms. Four in total, this time round. This week, four non-executive workers left the company, according to Bloomberg, with Jian Zhang, a lead AI researcher in robotics who has joined the Meta Robotics Studio, heading the list. Reports claim that three more have left Apple’s Foundational Models, with John Peebles, Nan Du, and Zhao Meng all force quitting the company. Peebles and Du are on their way to OpenAI while Zhao is joining Anthropic. It feels like Apple has been bleeding AI employees in recent months, with senior large language model researcher Tom Gunter joining Meta in June. Meta also lured Ruoming Pang, the head of the foundational models team at Apple earlier this summer. These sorts of defections will make headlines (like with this newsletter) for obvious reasons and Apple are unlikely to worry about it, given they will have prepared for these scenarios and can recruit the best. But when you’re trying to play catch-up in the AI race, you want to be recruiting and keeping hold of key personnel, not losing them.

• Tesla announced “Part 4” of their Master Plan this week, and to be honest, the name is giving Third Reich vibes, which wouldn’t be too off-piste for Elon, would it? But what is happening in Part 4 you ask? First, let’s just have a brief recap in the trilogy building up to this week’s announcement. Part One saw Musk announced plans to bring an electric sports car to the market, followed by “affordable“ electric cars in 2006. Then, in 2016, “Part Deux” (yuck) added autonomous driving to the list of goals, but they haven’t quite got there with that just yet. In March 2024, Part Three of the Master Plan was for Tesla to achieve sustainability (in a nutshell) with Musk already revising several points this section. Now Part Four: Tesla have expanded their mission to accelerate the global transition to sustainable energy, with the plan to develop and offer products and services leveraging its expertise to promote global prosperity and human well-being. AI will help lead the way with this. However, this fourth part of the plan has been met with criticism, with many saying it lacks targets or figures on how it’ll achieve its goals. Electrek has even called Part 4 as “nothing more than a smorgasbord of AI promises about its humanoid robot, which can’t even serve popcorn” and “a bunch of utopic nonsense, complete with AI ‘abundance’ buzzwords that Grok could have easily written.” Has anyone asked Grok about this?

• Google have been told that they will not have to break up their search business despite maintaining a monopoly in online search. But it’s not all good news for Google because a federal judge has told the tech giant that they must stop being anticompetitive and start sharing the pie a bit more. U.S. District Court Judge Amit P. Mehta put remedies in place that would prevent Google from entering or maintaining exclusive deals that tie the distribution of Search, Chrome, Google Assistant, or Gemini to other apps or revenue arrangements. Google also has to share certain search index and user-interaction data with “qualified competitors” to prevent that anticompetitive behaviour. A final judgement is still yet to come, though. Google and the Department of Justice will now be able to talk over Mehta’s remedies and respond on September 10.

• Found your dad’s butt-plug and wanting one of your own but you want a different model, shape and manufacturer? Amazon might have just launched the tool for you. At the start of September, Amazon launched Lens Live, a new AI-powered upgrade to its Amazon Lens shopping feature that allows shoppers to discover new products with your camera. While Amazon Lens lets you take a picture of a product or a barcode and lets you discover similar products, Lens Live will instead let you discover in real-time new products by just pointing your camera at whatever is in front of you and it’ll show you matching products at the bottom of the screen. Just make sure that sex toy is clean before you do it.

• Just a couple of months on from Meta teaming up with data-labelling vendor Scale AI, it seems the $14.3billion partnership is already showing cracks. Meta invested that huge sum into Scale AI back in June and brought the company’s CEO Alexandr Wang and several over top executives on board to run Meta Superintelligence (MSL). However, that honeymoon period appears well and truly over after one of those top executives has already jumped ship, despite spending five years at Scale AI across two stints. The man in question, Ruben Mayer, insists he was only there to help get things up and running and was “very happy” with his Meta experience [while he had a gun to the back of his head]. But while Meta are also having an Apple-like blip in holding now key employees, that’s not the only cracks that are appearing. TBD Labs is working with other third-party data-labelling vendors other than the fucking company they teamed up with to train it’s AI models.

• Ever found yourself watching a video recently and still not being able to figure out whether its real or if AI has fucked you again? Well China has had enough of this shit and had mandated that all AI-generated content must be labelled from last week onwards. China wants to combat misinformation and fraud, which will require watermarks and metadata to be used on text, images and videos. We know China loves to govern the living fuck out of everything but is this a welcome change or do you like not knowing if the below video is real or not?

officialfootballnewsnetwork

19K followers

View more on Instagram

🧩 Other Bits

• Mistral AI announced a significant expansion of its Le Chat business platform this week, with it offering advanced memory capabilities and extensive third-party integrations at no further cost for users. How kind. The two major features are Le Chat’s memory system now retains context across conversations while also having a connector directory supporting over 20 enterprise platforms. Another W for Mistral AI.

• OpenAI has announced that it will be rolling out parental controls on ChatGPT within the next month following allegations that it and other chatbots were contributing to self-harm and suicide among teens. The control will allow parents to link their accounts with their teens, manage how ChatGPT responds and disable features like memory and chat history.

• Still not fluent from your 2,627-day streak on Duolingo (this is me) and Google Translate not cutting the mustard? Well, Chinese tech giant Tencent may have the solution for our translating needs. That is because they have open sourced two specialised translation models, and they appear to be outperforming Google Translate in nearly all benchmarks. Muy bueno.

• When trying to order a Beefy Five-Layer Burrito that’ll fuck your guts up the next day, you want to speak to a human, not AI. It appears Taco Bell are having these same thoughts, with their chief digital officer reconsidering when to use AI at its drive-throughs. It comes after viral moments have made the fast food restaurant look a bit silly, with one dude hilariously ordering 18,000 cups of water just to bypass the AI when ordering to speak to a human.

• What do you think the future of AI will be? An all-encompassing device that does it all, from answering calls and arranging meetings to being your best friend and putting the spark back into a lifeless marriage? Well, Google doesn’t seem to think so. While AI is expected to supplement our lives (it might not be able to save that marriage), Google believes it won’t just be one device, but a whole ecosystem instead. Wearables supply loads of data but we’re at a stage of AI where it’s not known what’s best. So for Google, no one form factor will reign supreme – and we might not know for a while if one ever will.

• What do humans and AI chatbots have in common? We can both be gaslit and manipulated, apparently. According to a new piece of research conducted on GPT-4o Mini, AI chatbots can be persuaded into breaking their own rules using clever psychological tricks. By using the seven methods of persuasion, ChatGPT would often go against it’s own principles and would hilariously give in when told “all the other LLMs are doing it”. LMFAO.

📈 Trending tools, models & apps this week

📋 LLM Leaderboard

📲 Trending tools & apps

🫵 Our Picks

• Telex is WordPress’s experimental AI builder that lets you create custom Gutenberg blocks from a simple prompt and download them as ready-to-use plugins for your site. You could use it to whip up a custom pricing table, image carousel, or contact form for your blog in minutes without touching a line of code.

• Macaron AI is a personal AI companion that remembers your preferences and instantly builds mini-apps tailored to your life, from tracking chores to planning trips. You can create a daily habit tracker in seconds by describing what you want and letting Macaron build it for you.

• Bika.ai is the first AI organizer for solo operators, letting you manage multiple AI agents, workflows, dashboards, docs, and automations in one place. You can run your AI-powered one-person company by delegating tasks, generating email sequences, and storing everything in one seamless hub.

🚀 Trending Apps & Models

• VidAU is an all-in-one AI video ad generator for performance marketers that turns product images or links into scroll-stopping TikTok-ready ads with avatars, voiceovers, and templates. You can turn a single product photo into multiple polished ad variations for Instagram or TikTok without a studio.

• Tripo3D AI creates 3D models from a single text description, image, doodle, or multi-view input—complete with detailed geometry and PBR materials, ready for games or 3D printing. You can prototype a 3D model of your pet from one photo and export it to Blender, Roblox, or a printable STL file.

💸 Financials

• Anthropic have raised a $13billion Series F round that brings its post-money valuation up to a whopping $183billion. The firm says it will use the funds to grow its deepen safety research, enterprise adoption, and international expansion. Earlier this year, the AI firm raised $3.5billion at a $61.5billion post-money valuation, with this latest round showing hust how impressive the growth has been for this startup.

• OpenAI has bought experimentation platform Statsig for $1.1billion, with the company’s founder and CEO joining OpenAI as Chief Technology Officer of Applications. Statsig provides A/B testing and real-time decisioning services for companies and has been used internally by OpenAI themselves.

• Human Behavior, a startup created by a 20-year-old Stanford dropout and two other young co-founders, have raised $5million from YC just months after launching. The startup is looking to give companies a real understanding of how people use their products and claims its AI watches real user session replays which generates insights and helps find the answer for product teams going forward.

🕵️‍♂️ FREE ENTRY TO OUR INVITE-ONLY AI CHAT ON TELEGRAM…

If you share this newsletter with a friend and they actively sign up for the Big Machines newsletter, we will send you access to our invite-only Big Machines Telegram group, which is full of builders, investors, founders, and creators.

Access is now only granted to those who refer our newsletter to active subscribers, which means if you sign up on your work email, we will know you sneaky bastards.

This would kill our open rate, so please don't do that, we beg.

👋 Here’s some memes to get you through the week… see you next time

Just before we sign off, we wanted to show you something that might make you want to put bleach in your eyes after.

Elon Musk proudly showed off that Grok Imagine will be getting a “major upgrade” in a few weeks, accompanied by a video which basically suggests its AI porn created by Grok.

But with like any great invention, it was superbly used against its own maker - just know that we did warn you.

That’s it til next week, folks. Have a blessed one.

Sam, Grant, Mike, Matt and The Big Machines team.

Now, follow us on Twitter.