It’s Tuesday!
Let’s catch up on the latest in AI. Some big shifts are happening this week – from OpenAI’s cash flood to Apple quietly building its own ChatGPT. Let’s get into it 👇
TODAY IN AI
Claude quietly crushes student hackers
Anthropic’s Claude just scored in the top 3% at Carnegie Mellon’s PicoCTF hacking competition. A challenge made for humans, not bots. It had minimal help and still beat most student teams. The company says AI agents are now reaching expert-level cybersecurity skills. Attackers are leveling up, fast. Defenders need to catch up.
Source: AxiosOpenAI raises $8.3B and hits 5M ChatGPT business users
Another monster raise for OpenAI. Dragoneer led the $8.3 billion round, with backers like Blackstone, Sequoia, and Andreessen joining in.→ ChatGPT now has 5 million paid business users
→ Annual revenue is at $13B, with $20B in sight
→ The raise is part of a $40B funding effort still in motion
Source: CNBCAnthropic just blocked OpenAI from using Claude
Anthropic shut down OpenAI’s access to its Claude models. The reason? OpenAI allegedly used them for internal benchmarking, which violated Anthropic’s terms. OpenAI claims it’s standard practice, but Anthropic wasn’t having it.
Source: TechCrunchChina just built a monkey-brain AI computer
Researchers in China created a neuromorphic machine designed to mimic a macaque brain. Using 960 custom chips and 2 billion spiking neurons. They call it Darwin Monkey. It’s powered by DeepSeek and could bring us one step closer to brain-like AI.
Source: PCMagBP says AI is saving cash, might cut more jobs
BP is seeing big cost savings from AI, and it’s hinting at more layoffs. The oil giant already cut 8,000 jobs this year and says AI is now key to boosting margins across its legacy energy business.
Source: Sky News
PRESENTED BY SECTION
On September 25, get the expert formula for turning AI spend into ROI at your oganization at Section's virtual AI ROI conference.
No gatekeeping, no pitching, no BS. Hear from top AI leaders like Scott Galloway, Mo Gawdat, and May Habib and leave with a clear path to AI ROI for your company. Free to attend. Registration required.
Apple is quietly building its own ChatGPT
Turns out, Apple hasn’t just been playing middleman for ChatGPT. It’s building its own “answer engine” too.
The internal team, called AKI (Answers, Knowledge, Information), is working on a tool that could bring smart, conversational search to Siri, Spotlight, Safari – and maybe even a standalone app.
Here’s what we know:
→ Led by Robby Walker, former Siri boss
→ Hiring backend engineers for search
→ Still early, but moving fast
Why it matters: Siri still punts most queries to Google. But the $20B search deal with Google is under DOJ scrutiny, and Apple knows it needs options. Generative AI is shifting how people search, and Apple doesn’t want to be left behind.
Meta is hiring AI stars like it’s building the Avengers
Zuckerberg is on a hiring spree. Meta has recruited more than 50 top researchers for its new Superintelligence Lab, offering massive pay packages to talent from OpenAI, Apple, and beyond.
Recent hires include Shengjia Zhao (co-creator of ChatGPT) and Alexandr Wang. Meta reportedly offered ex-employee Andrew Tulloch up to $1.5 billion to come back. He passed.
But it’s not all smooth sailing. Some insiders say existing staff feel sidelined, and Sam Altman suggested Meta failed to land the elite talent they were aiming for.
Research backs that up. Harvard and HEC Paris studies show that too many stars on one team can hurt performance. Great teams need solid collaborators, not just rockstars.
Zuck’s bet is simple: more brains, more breakthroughs. But building a culture that actually works? That’s the hard part.
Trump cheers on AI companies in the copyright fight
Trump’s public support for AI companies didn’t make it into the White House’s AI Action Plan – but it still made waves.
Lobbyists from media, film, and music pushed hard to keep copyright out of the plan, and they got their wish. The White House left the issue out entirely, citing lack of consensus.
Then Trump got on stage and said it straight: “You can’t expect a successful AI program if you have to pay for every article or book you’ve read.”
AI companies loved it. One exec said they hope it sways future court decisions. Copyright holders? Not so much.
But legally, presidential opinions don’t mean much. Judges look at facts. Still, Trump’s remarks show how unpredictable AI policy is getting – especially when lobbyists and politicians have different playbooks.
OpenAI, Meta, and Nvidia are all ramping up visits to Mar-a-Lago and pouring money into lobbying. They’re betting big on a pro-AI, pro-business White House.
What did you think of this edition?
