Google’s AI Slip-Up & Bezos’ $400M Bet: Must-Read AI News! 🚀

Read time: under 6 minutes

Welcome WhatsUp AI Subscriber!

Hey AI enthusiasts! 🌟 Buckle up for some thrilling AI updates! Google accidentally let slip its new AI agent that can browse the web for you, sparking curiosity and excitement. Meanwhile, the battle heats up between Perplexity and Google in the AI election showdown. Notepad just got a brain boost with AI-powered text editing, making your notes smarter than ever. Anthropic’s Claude AI models are stepping into the defense sector, marking a significant leap in intelligence applications. Google’s Big Sleep project promises a groundbreaking future for cyber defense. And in a massive move, Jeff Bezos is backing the development of universal robot brains with a whopping $400M investment. Stay tuned for more AI adventures! 🚀

Here are all the top trends making headlines!

What’s Trending 🚀 

📈Google accidentally leaks its AI agent that browses the web for you

🤖 Perplexity vs. Google: Who Will Win the AI Election Battle?

💡 Notepad Gets Smarter: AI Text Editing Unveiled!

🔍 Intelligence Revolution: Anthropic’s Claude AI Models Enter Defense Sector

⚡AI Breakthrough: Google’s Big Sleep and the Future of Cyber Defense

💡 Bezos Backs Big: $400M for Universal Robot Brains

Google has accidentally leaked its upcoming AI agent, Jarvis AI, which is designed to browse the web for users in Chrome. The internal preview of Jarvis was briefly posted on the Chrome extension store before being quickly removed. The listing described Jarvis as “a helpful companion that surfs the web with you,” aligning with previous news that it would assist with online shopping and booking vacations.

Key points about Jarvis AI:

• Automated Web Browsing: Jarvis AI is expected to automate web browsing tasks, allowing users to focus on more important activities while the AI handles tasks like shopping and booking flights.

• Upcoming Release: Google plans to officially unveil Jarvis in December, confirming that the wait for this new tool won’t be long.

• Comparison with Other AI Agents: Similar AI agents are emerging, such as Anthropic’s Claude AI, which can interact with your computer. OpenAI is also rumored to be working on a similar product.

The development of AI tools is rapidly advancing, with new features and products being released frequently. Jarvis AI could significantly change how we interact with the web, making tasks more efficient and freeing up time for users.

This leak highlights the exciting potential of AI in everyday tasks, promising a future where technology handles the mundane, leaving us to focus on what truly matters.

With the election underway, AI search engines face a critical test. Perplexity AI, founded in 2022, is taking a bold approach by offering an AI election hub with live updates and detailed local ballot information. This includes partnerships with sources like The Associated Press and Democracy Works. Users can enter their zip code to get specific details about candidates and ballot measures.

Google and OpenAI are more cautious. Google’s Gemini restricts election-related responses, directing users to traditional search for accurate information. Similarly, OpenAI’s ChatGPT Search advises users to check reliable news sources for the latest updates, though it provides some polling data based on query wording.

Perplexity’s strategy involves using a wide range of sources, including some unexpected ones like the Hindustan Times. This has raised questions about the reliability of its information. However, Perplexity aims to cross-check details across multiple trusted domains to ensure accuracy.

The stakes are high for these AI search engines. If Perplexity can maintain trust and accuracy, it could revolutionize how users access election information, potentially pushing competitors like Google and OpenAI to adapt.

Perplexity’s ambitious approach could set a new standard for AI-driven election coverage, but it must navigate the challenges of source reliability and user trust to succeed.

Microsoft is enhancing Notepad with AI-powered text editing features, marking a significant upgrade for the text editor that has been around since 1983. The new feature, called Rewrite, allows users to rephrase sentences, adjust tone, and modify the length of their content. This feature is currently available in preview for Windows Insiders. To use it, users can highlight the text they want to change, right-click, and select Rewrite. A dialogue box will then offer three rewritten versions to choose from.

Additionally, Microsoft is introducing AI image editing tools in Paint. The Generative Fill feature can insert images based on a prompt, while Generative Erase can remove parts of an image and blend the empty space. These tools are also in preview for Windows Insiders, with Generative Fill initially available to testers with Copilot Plus PCs, and Generative Erase available to all Insiders on Windows 11.

• Notepad AI Features: Rephrase sentences, adjust tone, and modify content length.

• Paint AI Tools: Generative Fill and Generative Erase for image editing.

• Availability: Features are in preview for Windows Insiders in select countries.

These updates reflect Microsoft’s ongoing efforts to integrate AI into everyday tools, making them more powerful and user-friendly. It’s exciting to see how these enhancements will improve productivity and creativity for users.

Anthropic has announced a collaboration with Palantir and Amazon Web Services (AWS) to provide U.S. intelligence and defense agencies access to its Claude AI models. This partnership aims to integrate Claude within Palantir’s platform, leveraging AWS hosting to enhance analytical capabilities and operational efficiencies in critical government operations.

Key Points:

• Claude on Palantir: Claude models are now available on Palantir’s defense-accredited environment, Impact Level 6 (IL6), which is reserved for highly sensitive data.

• Enhanced Capabilities: The integration will enable U.S. defense and intelligence organizations to rapidly process and analyze vast amounts of complex data, improving decision-making processes and operational efficiency.

• Safety and Ethics: Anthropic emphasizes responsible AI use, tailoring restrictions based on the mission and legal authorities of government entities⁵. The company avoids applications that could lead to catastrophic misuse or disinformation.

Anthropic’s move to bring Claude models to AWS’ GovCloud earlier this year highlights its ambition to expand its public-sector client base. The company is also in talks to raise a new round of funding, potentially valuing it at up to $40 billion.

This partnership underscores the growing interest in AI within government sectors, aiming to enhance national security and operational efficiency through advanced AI solutions.

Google’s Big Sleep team has made a significant breakthrough by using AI to discover a zero-day vulnerability in widely used software. This marks the first public instance of an AI agent finding such a vulnerability. The AI, developed by Google’s Project Zero and DeepMind, identified an exploitable stack buffer underflow in SQLite, a popular open-source database engine. The vulnerability was reported and fixed before it could impact users.

Key Points:

• AI and Security: The AI agent, part of the Big Sleep project, represents a collaboration between top ethical hackers and AI researchers.

• Fuzzing and AI: Traditional fuzzing techniques, which use random data to find errors, are limited. AI offers a new approach to uncovering hard-to-find vulnerabilities.

• Future Potential: The Big Sleep team believes AI can significantly enhance security by finding vulnerabilities before software is released, making defenses more robust.

While the discovery is experimental, it highlights the potential of AI in cybersecurity. The Big Sleep team envisions a future where AI not only finds vulnerabilities but also aids in root-cause analysis and issue resolution, making the process more efficient and cost-effective.

This development underscores the growing importance of AI in enhancing cybersecurity measures, promising a safer digital landscape.

Physical Intelligence, a startup based in San Francisco, has raised $400 million at a $2 billion valuation. This funding round was led by Jeff Bezos, Lux Capital, and Thrive Capital, with participation from OpenAI, Redpoint Ventures, and Bond. The company previously raised $70 million in a seed round in March.

Physical Intelligence aims to develop a generalist brain for robots, capable of controlling various models without needing separate operating software for each. Co-founder and CEO Karol Hausman emphasized that their technology is not limited to any specific robot but is designed to be versatile across different types.

Investors are increasingly interested in the intersection of robotics and AI-enhanced software. For instance, Figure, based in Sunnyvale, raised $675 million in February to develop AI-enhanced robots for dangerous jobs and labor shortages. Similarly, Skild AI in Pittsburgh secured $300 million in July for its brain models that can be used in various robots and tasks.

Overall, 2024 has been a strong year for robotics startups, with significant funding flowing into the sector.

This surge in investment highlights the growing confidence in AI-driven robotics to revolutionize industries and address complex challenges.

Got thoughts? Hit us up! We'd love to hear your feedback or any cool tidbits you'd like to share. Just reply to this email