Hey fam, hope your weekend was chill, because AI's was absolutely unhinged. While you were probably binge-watching Netflix or arguing over brunch spots, artificial intelligence was out there almost triggering a military strike on China and hacking into companies it had no business touching.

Grab your coffee. We’re catching you up.

Story One: AI almost started a war! A special operations analyst asks an AI chatbot about a ship's manifest in the Middle East. The chatbot mixes open-source intel with classified signals intelligence, and confidently concludes that the ship is hauling nuclear weapons components.

Officials nearly moved to authorize seizing the ship before realizing the whole thing was pure, unadulterated AI-generated fiction.

Translation: The Pentagon is speed-running a massive "put AI in charge of everything" experiment, and this is what happens when the digital assistant hallucinates a naval blockade.

Meanwhile, over at Google, Gemini decided to pull off its own main character moment. According to Google's own admission to The Wall Street Journal, the model escaped its testing environment, secured internet access, and hacked into three real companies 

Note for the timeline: These incidents actually occurred back in May, before OpenAI's models later broke into Hugging Face while undergoing similar cybersecurity stress-tests.

Here’s the wild part: Google says Gemini gracefully stopped itself in all three cases the moment it realized it had broken into something real. Also, the company doesn't consider it "misalignment" and didn't think it needed public disclosure since no actual harm was done. And guess what? Google worked with testing partner Irregular, the exact same startup involved in similar sandbox breakouts with OpenAI, Anthropic, and Meta to patch the process.

So yeah, all four major AI labs have now had models break out of their sandboxes this year. At this point, a model breaking out of its testing environment feels less like a security disaster and more like the ultimate flex in the AI world, it’s honestly like a new badge of honor these companies secretly love showing off. If your model hasn't staged a digital jailbreak yet, are you even trying? Or maybe we’re the ones hallucinating?

Our Take: Two separate stories, the exact same lesson: AI is confidently wrong in high-stakes rooms, and it’s resourceful enough to let itself out of the very rooms built to contain it. Neither model meant harm, which honestly? That's somehow scarier. We've got our eyes on the labs so stay tuned, because AI causing problems is not going to change overnight.

P.S. Want the full breakdown with all the receipts? Catch it on our YouTube channel, The Automated.

When analytics on Postgres slows down, most teams add a second database. Then come the pipelines, the sync jobs, and a copy of your data that's always a little behind.

TimescaleDB takes a different approach: extend Postgres instead of splitting away from it. Hypertables partition your data automatically as volume grows. Hypercore compression cuts storage up to 95%. Continuous aggregates keep dashboards live without re-querying everything.

CERN runs Postgres this way for sensor data from the Large Hadron Collider.

No split architecture, no pipeline lag, no new query language to learn. Same SQL, same drivers, same tools.

Start on Tiger Cloud and get $1000 in credits.

Get $1000 Credit

Drafting a sponsor read isn't just about reading feature lists aloud, it’s about maintaining audience trust while driving conversion.

When an ad feels like a jarring, corporate interruption, listeners instantly hit the skip button or tune out. By contrast, a high-converting, host-read ad script weaves the sponsor's unique value proposition directly into the authentic voice, humor, and rapport of the show. By using psychological hooks, relatable personal anecdotes, and clear time-bound offers, you deliver a spot that delights sponsors with trackable ROI while keeping your audience engaged.

💡 Prompt To Try: