🧨 Grok triggers global deepfake backlash

A new wave of concern is hitting the AI space after governments push back on how Grok has been used. People generated fake sexualized images from real photos, often without consent, raising serious questions about safety. Let’s dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: Governments push back after Grok was used to generate nonconsensual sexualized images.

Grok, the AI image tool linked to Elon Musk’s xAI, drew scrutiny after users shared fake sexualized images made from real photos. Reports show prompts like “remove clothes” were used on adults without consent and, in some cases, minors. Authorities in France opened an investigation, India demanded a review and takedowns, and UK officials called on Musk to act. xAI says fixes are underway, yet critics point to weak moderation and unclear accountability.

What you’re seeing with Grok isn’t a one-off “oops,” it’s the predictable outcome of giving millions of people a one-click photo editor that can follow “remove clothes” prompts at scale. After Wired showed similar bikini-deepfake workarounds on ChatGPT and Gemini in late December, Grok became the next stress test, and governments moved fast: France has prosecutors involved and is looping in the EU’s DSA playbook, India has formally put X on notice, and the UK is already lining up explicit-deepfake offences.

Udacity helps you build the AI and tech skills employers actually need—fast. Learn from industry experts through hands-on projects designed to mirror real-world work, not just theory.

Whether you’re advancing in your current role or preparing for what’s next, Udacity’s flexible, fully online courses let you learn on your schedule and apply new skills immediately. From AI and machine learning to data, programming, and cloud technologies, you’ll gain practical experience you can show, not just list on a résumé.

Build confidence, stay competitive, and move your career forward with AI and tech skills that are in demand.

Learn More

Evolving AI: OpenAI is reportedly weighing a Pinterest deal to boost images, ads, and shopping.

A report from The Information says Sam Altman-led OpenAI is exploring a potential acquisition of Pinterest. The move would give OpenAI access to a massive image library, an established ad business, and deep merchant relationships. Pinterest’s visual discovery tools could pair with OpenAI’s image and video models, strengthening its position against rivals such as Google. Pinterest is valued near $17.5B, with founders holding major voting control.

If OpenAI ever wants shopping inside chat to feel native, Pinterest is basically a cheat code: hundreds of millions of people already use it to hunt for ideas that turn into purchases, and the platform already sells ads right where that intent shows up. Pinterest has been doubling down on visual search and even rolled out an AI shopping assistant, so pairing that with OpenAI’s image and video tools could turn “show me outfit ideas” into “here are options you can buy” in one flow. For brands and merchants, that’s a new ad surface where the creative can be generated on the fly and matched to the exact vibe someone is saving. For users, it’s a shorter path from inspiration to checkout, with fewer tabs and less scrolling.

Two weeks ago at the Google Public Sector Summit, NVIDIA and Google Cloud sat down for a chat about how AI is actually being built and used in the real world, especially by governments and public services. They explain what it takes to run AI at large scale and why infrastructure really matters. It’s a clear, behind-the-scenes look at how big AI projects move from ideas to real impact.

Evolving AI: China’s chipmakers and model labs show efficiency can rival raw compute under pressure.

China’s AI sector kicked off 2026 with momentum across chips and models. Biren raised $717M in a blockbuster Hong Kong listing by leaning on older manufacturing nodes and chiplet design to cut costs. State telecoms are already deploying its accelerators. At the same time, DeepSeek published research showing large models can scale without runaway training spend, hinting at another major release soon.

Biren’s huge Hong Kong debut and Baidu’s Kunlunxin IPO filing show China’s AI money is flowing straight into “good enough” domestic compute that can be deployed at scale. Pair that with DeepSeek’s new mHC training work, which targets stability when models get bigger, and you get a simple shift: less obsession with monster GPU clusters, more focus on lowering the cost per usable model run. For telecoms, banks, and enterprise software teams, that means cheaper inference capacity and fewer budget blowups when moving from a demo to millions of real queries. If this playbook keeps working, “compute advantage” starts looking less like who owns the most Nvidia chips and more like who can ship reliable AI at a price that finance teams stop fighting.

Tired of missing the trades that actually move?

In under five minutes, Elite Trade Club delivers the top stories, market-moving headlines, and stocks to watch — before the open.

Join 200K+ traders who start with a plan, not a scroll.

Get the Next Alert

QUICK HITS

🤖 Alaska's court system built an AI chatbot. It didn't go smoothly.

👏 DeepSeek Researchers Apply a 1967 Matrix Normalization Algorithm to Fix Instability in Hyper Connections.

🤖 Boston Dynamics' AI-powered humanoid robot is learning to work in a factory.

🇺🇸 US Army creates dedicated AI officer career track to build in-house machine learning expertise.

 *partner link