🪪🛂 Claude Might Ask for Your Passport Starting July 8

Hey Explorers!

AI learned to sit in your Slack and crank out a finished 30-second ad in one take this week. In the same few days, it started asking to see your ID. The smarter it gets, the more everyone wants proof there's a human somewhere in the mix.

Here's the signal that matters 👇🏼

🤝 Hand Off a Task in Slack, Go to Lunch, Come Back to It Done
Claude now sits inside Slack as a teammate you can tag. Drop a task on it and it pulls context straight from the channel, works the problem for hours while you're off doing literally anything else, and even chases down the threads that died because everyone got too busy to follow up. The whole channel shares one Claude, so your team picks up where it left off instead of re-explaining the same thing five times. (🔗 Link)

🪤 Every Time a CEO Predicts AI Will Kill a Job, Hiring for It Climbs
Dario Amodei said engineers were 6 to 12 months from extinction. Demand for them hit record highs instead. Sam Altman called time on customer support, and support hiring took off. Kevin Indig's read: pitching your product as a people-replacer is the cardinal sin of AI marketing, because it wins the headline but quietly rots the trust you need with the exact buyers and employees you're talking to. Position it as a power-up for your team. The replacement pitch is the one that backfires. (🔗 Link)

🦁 Cannes Stopped Asking If You Used AI. Now It Asks If You Mattered.
For 70 years Cannes gave trophies to the ad. This year it started giving them to proof that a human shaped the thing. The new AI Craft category wants disclosure, provenance, and evidence a person made the real calls instead of just running slick prompts. Cranking out a thousand options costs nothing now. The taste to know which one's good is the whole job. (🔗 Link)

📝 Wikipedia Built the Cheat Sheet for Catching AI Copy. Read It Before You Hit Publish.
Wikipedia's editors waded through thousands of AI-written submissions and turned the patterns into a public list: the rule-of-three stacking, the "it's not X, it's Y" flips, the em-dash addiction, the hollow "stands as a testament" filler. They built it to guard the encyclopedia, but it reads like an autopsy of every B2B email in your inbox. If a stranger can spot the bot in your copy, your customer can too. Run your drafts through it first. (🔗 Link)

🎬 Feed It Your Brand Kit, Get Back a Finished 30-Second Ad
ByteDance's Seedance 2.5 makes one unbroken 30-second clip at 4K, and it'll take 50 references at once, so your whole brand kit goes in and a full spot comes out the other side. The enterprise video arm is already pulling $2 billion a year, with CapCut's 400 million users sitting right downstream. The catch that should give you pause: every major Hollywood studio hit the last version with cease-and-desist letters, and that fight is still going. (🔗 Link)

🚦 Washington Now Picks Who Gets OpenAI's Next Model, One Customer at a Time
GPT 5.6 isn't going public on launch day. The White House told OpenAI to hand it to a vetted list first and clear access customer by customer, with everyone else maybe a couple weeks back. That's twice in a month the government's tapped the brakes on a big release. Whether you get to run the most powerful models is starting to look like Washington's call, not OpenAI's. (🔗 Link)

🎲 Meta's Building an App Where the AI Writes the Bet and Also Decides If You Won
Leaked docs show Zuckerberg has a team on Arena, a prediction market where Llama dreams up the questions and then rules on who was right. Same AI on both sides of the wager. It runs on play money for now, which happens to buy Meta time to sort the gambling licenses while everyone gets comfy with AI calling balls and strikes on what's true. Neat trick. Mildly horrifying. (🔗 Link)

🪪 Come July 8, Claude Might Ask to See Your Passport
Anthropic quietly changed its policy so Claude can, in some cases, make you verify your age or identity with a real government ID, run through a third party called Persona. Supposedly only a small slice of users will hit it, mostly around fraud and misuse. Discord, Reddit, and ChatGPT already walked this path, so you can see where it goes. Just opening a chatbot and going is ending. Your ID is becoming part of the login. (🔗 Link)

Delegating to Claude in Slack becomes normal faster than you'd think. So does the overnight ad render. What doesn't get easier is proving a human was in the loop when a buyer, a regulator, or an awards jury comes asking.

Everyone's racing to bolt on what AI can do. Almost nobody can answer the question coming right behind it: where's the human, and can you prove it?

At Purple Horizons, we help your team put AI to work with the human judgment and checkpoints that hold up when someone asks you to prove it.

👉🏼 Work with Purple Horizons

Signal over noise. Always.

Ralph & Gianni

Most teams scale customer service by hiring, but with ElevenAgents, real-time voice and chat agents handle high volumes of interactions and resolve them faster.

When routine requests come in, they get answered in the moment, in 70+ languages. However, the complex ones route to humans with full context, so your team focuses only on the conversations that need judgment. The experience holds up because the agents sound human, not robotic. And with Expressive Mode, they read tone and adapt - de-escalating, reassuring, and guiding each conversation to a clear resolution.

You can train every agent on your knowledge base, SOPs, and policies, deploy across voice, chat, and WhatsApp, then keep improving with real-time CSAT tracking, A/B testing, and Guardrails.

The outcome is the one leaders care about: higher resolution rates, higher CSAT, and support that scales with demand instead of headcount. Pricing is transparent and flat at $0.08 per minute, with no add-ons to work around.

Get started free

📍 7/7 South Miami Tech Tuesdays @ Bougies

📍 7/14 Cafecito & Pastelitos @ maman West Palm Beach

Want to sponsor our newsletter or events? Drop us an email at [email protected]