⚖️ OpenAI accuses DeepSeek of copying U.S. AI models

🤖 AI bot swarm on X spreads crypto scams

🏠 Airbnb says AI now handles one-third of support tickets

👓 Meta considers facial recognition for smart glasses

🧺 $8,000 robot promises to fold your laundry

+ 📊 Daily poll and results

+ 📈 Trending tools and resources

OpenAI told a U.S. House committee that Chinese startup DeepSeek is attempting to “free-ride” on American AI research, alleging it used distillation techniques and workarounds to replicate capabilities from U.S. frontier models.

The accusations come as DeepSeek prepares for a possible new model announcement, after its R1 release last year claimed near top-tier performance despite limited access to advanced chips.

The dispute highlights broader U.S.-China AI tensions, with China pushing open-weight models and U.S. firms favoring closed systems, as both sides compete for leadership in the global AI race.

Researchers found over 1,000 AI-powered bot accounts on X running crypto scams, using realistic conversations and coordinated engagement to trick the platform’s algorithm and gain influence.

New AI models make these bot swarms more advanced, able to tailor messages, blend in with real users, and create the illusion that “everyone agrees,” a tactic known as synthetic consensus.

Experts warn this threatens elections and public debate, and say solutions like better detection tools, AI watermarking, and more access to platform data are needed to limit the damage.

Airbnb revealed that its custom AI agent now resolves about one-third of customer support cases in the U.S. and Canada, with plans to expand globally and eventually handle over 30% of support across all supported languages.

CEO Brian Chesky said the shift will significantly cut costs and improve service quality, as the company builds a more “AI-native” experience under new CTO Ahmad Al-Dahle, formerly of Meta’s Llama team.

Airbnb is also integrating AI into search and trip planning, while emphasizing its proprietary data, payments infrastructure, and host network as competitive advantages against emerging AI platforms.

Meta is reportedly developing a facial recognition feature called “Name Tag” for its Ray-Ban smart glasses that could identify people and surface information about them through its AI assistant, potentially launching as soon as this year.

Internal discussions suggest the tool would not allow users to identify just anyone, but may recognize contacts or public social media profiles, raising privacy and civil liberty concerns given Meta’s past retreat from facial recognition on Facebook.

The move comes as Meta’s smart glasses gain traction, with millions sold, and as competitors like Apple and Google prepare their own AI-powered eyewear, intensifying the race to define the next generation of wearable tech.

Bay Area startup Weave Robotics opened pre-orders for Isaac 0, a stationary home robot designed to fold laundry, priced at about $8,000.

The robot takes 30 to 90 minutes to fold a load and still struggles with large blankets, bed sheets, and inside-out clothes, with human teleoperators assisting on more complex items.

Despite limitations and a price comparable to a used car, Isaac 0 signals progress toward bringing advanced household robotics into everyday homes.

A. Yes, like handwritten letters - 73% 🏆

B. No, it’ll always have value - 27%