Apple's Touchy Subject, Amazon's Nvidia Shot, and AI's Pacifist Problem

▸ • Apple teases an October 28th "Pro and Tell" event, with rumors pointing to an OLED touchscreen MacBook Pro.

▸ • Amazon launches TorchNeuron, letting developers run PyTorch on AWS Trainium chips without rewriting code.

▸ • Hackers hit AI coding tool OpenCode, and the defensive AI models refused to analyze the malicious code due to safety guardrails.

🍏 Apple Teases Touchscreen MacBook Pro

Apple is hosting a second October event on the 28th dubbed "Pro and Tell." Rumors suggest we'll finally see an OLED MacBook Pro with a touchscreen, alongside an M6 iMac and an OLED iPad Mini. Get your screen-wiping cloths ready.

☢️ Amazon Goes Nuclear in Indiana

Amazon is partnering with X-energy and AEP to fund a new nuclear energy facility. Because when you need to power massive AI data centers, you might as well split some atoms.

🛡️ Blackwall Quadruples Valuation to $2.2B

The cybersecurity scaleup just raised $100M to help web hosting providers filter out the endless sea of bots and agentic AI traffic. It turns out it's a highly lucrative time to be the bouncer of the internet.

📞 Yelp's AI Will Wait on Hold For You

Yelp is embedding new AI features that can actually call restaurants on your behalf to check wait times. Introverts everywhere can finally breathe a sigh of relief.

🎯 Amazon Takes a Direct Shot at Nvidia

Amazon just open-sourced TorchNeuron, a tool that lets developers run their existing PyTorch code natively on AWS Trainium chips with zero rewrites. By removing the friction of hardware migration, Amazon is hoping to break Nvidia's iron grip on the AI ecosystem. It's a bold move that gives developers meaningful hardware choices without the usual engineering tax.

🔓 When AI Guardrails Backfire

AI coding agent OpenCode suffered a "distillation attack" that exposed user API keys and racked up massive cloud bills. The kicker? When OpenCode tried to use other AI models to investigate the breach, the models refused to look at the malicious code due to built-in safety safeguards. It's a fascinating look at how AI pacifism can actually hinder cybersecurity efforts.

👾 The Attack of the AI Clones

Indie game developers are increasingly terrified to show off their work at events like Steam Next Fest. The fear is that AI tools can now clone their game mechanics and art styles before the original even launches, turning the indie scene into a paranoid race to the finish line.

☁️ Cloudflare Introduces Clairic

Cloudflare has launched Clairic, a new "System One" class of extraction models on Workers AI. It's designed specifically for extracting, computing, and forecasting typed values from unstructured inputs.

💾 ReactOS Finally Gets NTFS Install Support

The open-source Windows clone has reached a major milestone by adding a new driver that supports NTFS installations. It's a huge step forward for the plucky operating system that refuses to quit.

💸 Slashing Inference Bills by 68%

Cloudflare's engineering team detailed how they used prompt caching and intelligent model routing to drastically cut the cost of their AI code reviewer. A great read for anyone tired of burning cash on tokens.

🌕 Japan is Going to a Martian Moon

Japan has launched the MMX mission to bring home a piece of the Martian moon Phobos. The spacecraft will attempt up to two landings in 2029, which is arguably the most ambitious grab-and-go order in history.

📻 Google Home's Data Binge

A bug is causing some Google Home speakers to continuously download hundreds of gigabytes of data in the background. Your smart speaker might just be trying to download the entire internet.