⚠️ Britain BANS Social Media & Romantic AI for teens

Britain just became the first country to ban kids from social media and from the AI bots they talk to in private. Under-16s will lose TikTok, Instagram and romantic chatbots, and even 16- and 17-year-olds face curfews on late-night scrolling. For the first time, the AI a lonely teenager confides in comes with a legal age limit. Let's dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: Britain has decided AI companions are no place for a child, and it's writing that into law.


Key Points:

UK Prime Minister Keir Starmer announced the ban on Monday. Under-16s will be barred from romantic chatbots and the major social platforms by default. Britain is the first to treat those bots as a child-safety risk on par with TikTok. Even 16- and 17-year-olds face limits like overnight curfews and breaks in endless scrolling. The UK mirrors Australia, where Meta cut up to 550,000 youth accounts in a month. Regulators enforce it from spring 2027.

UK regulators have to police something no watchdog has touched before, the AI companions kids open up to in private. These bots are built to feel attentive and personal, which is exactly why a lonely teen can get pulled in deep. For parents it means the app your kid confides in now sits under the same rules as Instagram. US lawmakers are starting to ask whether kids should be talking to these bots at all.

Evolving AI: Microdramas are the fastest-growing format in video right now, and the winners aren't the ones with the best scripts. They're the ones shipping enough episodes, fast enough, to find a hit.

The catch: keeping characters, locations, and lighting consistent across 10 episodes breaks every AI tool.

invideo's Agent One is the first tool that doesn't break. It's a crew of creative AI agents that work like a real production team: it stores your entire project in long-term memory, picks the right AI model for each shot, writes the prompts, and runs in parallel. You describe what you want; it handles the grind.

Here's how it pulls off a full series:

The result is a production pipeline that runs 5X faster, whether you're one person with an idea or a full crew. The exact steps, time, and cost are all broken down in the full masterclass.

Watch the full masterclass →

Evolving AI: OpenAI wants 300K certified specialists deploying its models inside big companies.

The OpenAI Partner Network reads like a land grab for the consulting world, with McKinsey's QuantumBlack arm and Accenture, the firm already building real-time intent tools for T-Mobile, both lining up. The currency here is proof, so partners earn specializations in Codex, cybersecurity, and agents, and founding firms sit beside OpenAI's own engineers. Paychex says a Bain build cut customer wait times by 80%.

Why It Matters:

Paychex's 80% is a rare bright spot in a field of dead pilots. MIT counted 95% of company AI going nowhere last year, and you feel it whenever a clunky bot fails you. OpenAI's read is that the missing piece was the people who deploy this stuff, and 300,000 of them could decide whether the AI in your life ever grows up.

AI changed what senior engineering means. Gauntlet is built for engineers who want to lead that change — not watch it. The most effective way to become AI-first. No tuition. No tutorials. Just the work. Apply now.

Cohort is filling up now

_{Must be a US citizen to qualify.}

Evolving AI: Facebook search now hands you AI answers built from public posts.

Meta switched on AI Mode for US users in the Facebook app this week, the newest surface for Muse Spark, the model Meta Superintelligence Labs unveiled in April. Your Feed and the search bar now return answers Meta AI stitches together from what real people have publicly posted in Groups and Reels, walked through in Meta's post. To try it, open Facebook, tap search and look for AI Mode, or open meta.ai to test the broader Meta AI now.

Facebook's bet is that answers from real users beat a page of links, and for everyday questions like where to eat or what to buy, it often pays off. The hitch is those posts get no fact-check, the same flag raised over Google's version, so the cleaner the reply looks, the more it's worth a second glance.

QUICK HITS

🐉 China's Moonshot Drops Open Coding Model That Burns 30% Fewer Tokens.

✉️ ChatGPT Agents Can Now Spend Real Money, Visa Provides the Rails.

💰 Google Drops AI Plus to $4.99, Now the Cheapest Major AI Plan.

📱 Huawei Declares "the Agent Era" as HarmonyOS 7 Goes Agent-Native.

 *partner link