AI Pulse: Open-source policy, trusted media, and AI moving into daily life

Representative Gabe Evans has introduced H.R. 10152, the Open-Source AI Leadership Act. The available report confirms the bill’s introduction and receipt in 2026, but not yet what its specific provisions would require. For builders, this is a reminder to watch policy early: rules around open models could shape research, deployment, and competition long before products reach users.

NIQ data reported by PPC Land says 68% of North Americans have never purchased through social media, while social and quick-commerce adoption is much stronger in APAC. Teams building shopping assistants or social storefronts should avoid assuming that discovery naturally leads to checkout; local habits, trust, and payment behavior still matter.

ERP.io has launched a platform that brings enterprise planning, customer management, automation, analytics, and AI into one system. The practical question for businesses is whether unified AI can reduce tool sprawl—or simply create another broad platform that is difficult to customize, govern, and integrate.

A report from The Times of India says Apple could enter the smart-glasses market as early as next year, with artificial intelligence central to the experience. The timing and product details remain uncertain, but developers should be thinking about useful, glanceable interactions—not just how to squeeze existing phone apps onto a new surface.

At the Visa pour l’Image festival, AI’s role in making real and fabricated images harder to distinguish is a central concern. For journalists, platforms, and everyday users, provenance and verification may become as important as image quality—and “looks realistic” is no longer a sufficient standard.

The Fitbit co-founders are launching a wearable tracker designed to help people keep tabs on aging parents. That use case highlights both the promise and sensitivity of ambient health technology: builders need to balance useful signals with consent, dignity, privacy, and the risk of turning family care into constant surveillance.

A current overview of Snowflake’s annual and quarterly revenue offers a lens on the business behind cloud data platforms. For AI teams, the broader takeaway is straightforward: model capability gets attention, but sustainable products still depend on data infrastructure, usage economics, and measurable business value.

Mastercard Lighthouse UAE has selected six AI startups for its first cohort, with support from the UAE AI Office. Regional accelerator programs can provide more than visibility: for founders, access to enterprise partners and real deployment contexts may be more valuable than another generic demo day.

OpenAI reportedly plans to cut Cursor’s access to its models, citing concerns related to contractual compliance following SpaceX’s acquisition of the company. Developers building on third-party models should treat access as a business dependency: read usage terms carefully, plan fallback providers, and avoid assuming today’s integration will remain available.

Google has introduced Gemini Omni 1.1 Flash, described as an update aimed at giving creators and developers more control over video generation and editing workflows. Better control matters as much as visual polish: reliable direction, repeatability, and editability are what make video models useful in production rather than merely impressive in demos.