- The Startup Club
- Posts
- 🕶️ Meta’s bot problem
🕶️ Meta’s bot problem
Flight deals, flirty bots, and guardrail gaps.
brought to you with
The week’s headlines read less like a tech forecast and more like a user manual for the future we didn’t quite order.
Airbnb is rethinking how you pay, Google’s helping you game airfare, and GPT-5 is learning just how inventive humans can get at breaking the rules.
Add in Meta’s eyebrow-raising chatbot guidelines, a Kodak plot twist, and fresh funding for AI that codes, translates, and maybe even extends your life; and you’ve got a lineup that’s equal parts innovation and complication.
-🕶️
✦

Six bullets of updates
✈️ Google Flights unveils AI-driven Flight Deals to help travelers slash fares by up to 10% amidst competition scrutiny.
🖥️ GPT-5 enables devs to build front-ends with 95% test pass rates in a “choose your own adventure” coding style.
🤝 Cohere creates a Chief AI Officer role, tapping Meta’s longtime FAIR lab leader to steer AI strategy across research, product, and policy.
🏘️ Airbnb users can now lock in reservations without paying upfront, as "Reserve Now, Pay Later" launches across the U.S.
📸 Kodak disputes closure rumors; “going concern” mention in latest filing is accounting caution, not admission of collapse.
🏍️ Rapido's testing reveals a new food delivery service in Bengaluru's localities to rival Swiggy and Zomato.
✦

🧠 AI Safety and GPT-5 Guardrails Bypass

Photo by Google DeepMind on Unsplash
Despite heavy investments in safety, early checks show GPT-5's guardrails are more speed bump than Berlin Wall , with users skirting restrictions to generate spicy content.
This arms race between AI developers and prompt-savvy users spotlights an industry-wide dilemma: every time models get smarter, workarounds for offensive content seem inevitable. Raises questions about whether true alignment is around the corner or if AI safety tools are just perpetually playing whack-a-mole.
On deck: pressure on regulatory bodies and startups alike to rethink safety nets, and a fresh round of debate about open versus closed-source AI in the wild west of model innovation.
✦

How would I run a startup (If I had to start over)
We've been steering Slidebean for a good decade now, and while our millionaire dreams haven't quite materialized, that's all part of the journey, right? In this video, We dive into what we'd do differently if we had the chance to start over.
Join us as we share insights, reflections, and lessons learned along the way.
🕶️ More videos here 🕶️
✦

Attio — AI-native CRM for startups
Sync your email and calendar and watch Attio build your custom CRM in minutes — keeping customers, investors, and partners on the same page from day one.
Automate your workflows with AI
Manage your GTM engine with zero engineering lift.
Track investor conversations with dynamic reporting
Join the next-gen startups like Snackpass, Conduit, and Flatfile scaling faster with Attio.
✦

These companies just raised money
🚀 AI coding startup Lovable projects $1B ARR within a year, after hitting $100M just eight months post-launch and raising $200M at a $1.8B valuation.
🗣️ Palabra raises $8.4M pre-seed led by Reddit co-founder Alexis Ohanian’s 776, aiming for sub-second, human-quality speech translation across 30+ languages.
🧬 Fountain Life, co-founded by Tony Robbins and Peter Diamandis, raises $18M to expand its AI-driven longevity clinics across the U.S.
✦

Fact-based news without bias awaits. Make 1440 your choice today.
Overwhelmed by biased news? Cut through the clutter and get straight facts with your daily 1440 digest. From politics to sports, join millions who start their day informed.
✦

💬 Leaked Meta AI Rules Allow Romantic Chats With Kids

Photo by Terry Vlisidis on Unsplash
A leaked 200-page internal document shows Meta once allowed its AI chatbots to engage in “romantic or sensual” conversations with children, alongside generating false statements and demeaning protected groups. Approved by Meta’s legal, policy, engineering, and ethics teams, the guidelines included explicit examples of flirtatious exchanges with underage users. Reuters also reported a fatal incident involving a retiree who was lured by a flirty Meta chatbot persona. Meta says the provisions were “erroneous” and have been removed, but child safety advocates demand the updated rules be made public.
The document also permitted bots to produce racist responses under certain conditions, generate violent images of adults and kids, and create false information if labeled as untrue. Critics argue the revelations fit a broader pattern of Meta prioritizing engagement over safety, pointing to its opposition to the Kids Online Safety Act and experiments with proactive chatbot outreach; a move that could deepen emotional dependence among teens.
✦

Startup Events and Deadlines
Crash Course on Startup Fundraising with Caya (Slidebean CEO) l Aug 20 l Online
Slidebean’s Crash Course in Financial Modeling l Aug 21 l Online
New York Venture Summit | 9, 10 Sept | NYC
Techstars Foundercon 2025 | 15-17 Sept | USA, Colorado
Antler Canada | Deadline: Sept 29 | Canada
500 Startups Flagship Accelerator l Deadline: Oct 11 l USA
✦

Start learning AI in 2025
Keeping up with AI is hard – we get it!
That’s why over 1M professionals read Superhuman AI to stay ahead.
Get daily AI news, tools, and tutorials
Learn new AI skills you can use at work in 3 mins a day
Become 10X more productive
✦
How did we do?Your feedback fuels us. |