🔥Baidu vs. DeepSeek Heats Up

Today we will discuss:

• Baidu’s AI push—The Chinese tech giant has launched Ernie 4.5 & X1, but DeepSeek’s low-cost models are raising the stakes

• Google’s AI sparks controversy—Gemini 2.0 Flash removes watermarks, raising copyright concerns

• In other AI news today—Google integrates its HD voice model Chirp 3 into Vertex AI, OpenAI and Musk agree to a fast-tracked trial over OpenAI’s for-profit shift, the Russo Brothers launch an AI-powered studio for films and games, AI nurses face pushback from human nurses over patient care concerns

• Don’t miss this week’s Workflow Wednesday—AI and creativity unlocked with Sudowrite.com, Suno.com, and Descript.com; now available at only $20/month! 

• Saved the best for last—10 must-try AI tools

Baidu unveils new AI models as DeepSeek gains ground

Key Points 

• Baidu has launched Ernie 4.5 and Ernie X1, with claims of high performance and strong multimodal capabilities.

• While Baidu was an early AI leader, rising competition from DeepSeek has pushed the company to accelerate its AI roadmap, with Ernie 5 set to launch later this year.

☕News – Baidu has introduced two new AI models: Ernie 4.5, an upgraded version of its foundational model, and Ernie X1, a reasoning model designed to enhance complex problem-solving.

The company claims that Ernie X1 delivers performance comparable to DeepSeek R1 but at half the cost. Ernie 4.5, meanwhile, is being promoted for its "high EQ," with the ability to understand memes and satire. Both models support multimodal capabilities, meaning they can process video, images, and audio alongside text.

🥸Baidu’s big problem – Despite being one of the first Chinese companies to launch a ChatGPT rival, Baidu has struggled to gain widespread traction. At the same time, DeepSeek has drawn attention by releasing competitive models at significantly lower prices, raising concerns among U.S. AI companies and investors.

In response, Baidu is accelerating its AI roadmap, with plans to release Ernie 5 later this year, promising even more advanced multimodal capabilities.

Read the full story here. 

Google’s new Gemini AI model sparks controversy over its ability to remove watermarks

Key Points 

• Unlike some AI models, Gemini 2.0 Flash doesn’t block watermark removal and even reconstructs missing image details seamlessly.

• While labeled “experimental,” the tool is available in developer platforms, making its potential misuse a concern for copyright holders.

♨️News – Social media users have discovered a controversial capability in Google’s new Gemini AI model: the ability to remove watermarks from images, including those from stock media companies like Getty Images.  

🧐For context – Last week, Google expanded access to Gemini 2.0 Flash’s image generation and editing tools. While the model is powerful, it appears to have few safeguards in place. It can generate images of celebrities and copyrighted characters without restrictions—and as users have demonstrated, it can also remove watermarks from existing images.

What makes this particularly concerning is that Gemini 2.0 Flash doesn’t just erase a watermark; it attempts to fill in the missing parts of an image, making the edits harder to detect. Other AI tools have similar capabilities, but this model stands out for its effectiveness—and the fact that it’s free to use. That said, it isn’t perfect. It struggles with semi-transparent watermarks and those covering large portions of an image.

😵‍💫Why this is problematic – Google currently labels Gemini 2.0 Flash’s image feature as “experimental” and “not for production use,” and it’s only available in developer tools like AI Studio. Still, copyright holders are likely to push back. Under U.S. law, removing a watermark without permission is generally illegal, with few exceptions.

Also, by comparison, other AI models, including Anthropic’s Claude 3.7 Sonnet and OpenAI’s GPT-4o, refuse to remove watermarks entirely. Claude explicitly states that doing so is “unethical and potentially illegal.”

Read the full story here. 

We’ve just launched the Tenth edition of Workflow Wednesday for AI-minded professionals like you—actionable AI workflows delivered directly to your inbox.

This week’s topic: AI-Powered Creativity

Join us here