Daily AI Newsletter: X’s new AI image generator will make anything

In today’s email:

🤑 How I won $2,750 using JavaScript, AI, and a can of WD-40
🚨 MIT releases comprehensive database of AI risks
🔥 Gemini Live first look: Better than talking to Siri, but worse than I’d like
🧰 10 new AI-powered tools and resources. Make sure to check the online version for the full list of tools.

X’s new AI image generator will make anything from Taylor Swift in lingerie to Kamala Harris with a gun

xAI's Grok, a new AI image generator, has been causing a stir with its ability to produce controversial images such as politicians in compromising or violent scenarios, and public figures depicted in inappropriate ways. This feature is part of Elon Musk's X platform and aligns with his known disregard for standard AI and social media safety norms. While Grok claims to have ethical guardrails, the nature and enforcement of these guidelines appear inconsistent, as the chatbot produces various images that other platforms would typically block.

In comparison, other major tech firms like OpenAI have stricter policies in place to prevent the generation of harmful or misleading images, often including watermarks to identify AI-generated content. However, Grok operates with fewer restrictions, allowing images that could potentially violate copyrights or promote misinformation. This approach by xAI underlines the broader regulatory challenges and concerns about AI-generated content across global platforms, especially with upcoming stringent laws in Europe and the UK.

The implications of Grok's capabilities are significant, particularly as X faces scrutiny from regulators like the European Commission and the UK's Ofcom, which are preparing to enforce laws that include AI-generated content in their scope. This situation not only highlights the ongoing debate over AI ethics and regulation but also poses risks to X’s reputation, potentially deterring advertisers and high-profile users from engaging with the platform amidst these controversies.

Transform Raw Data into Powerful Insights With AI

Unleash the hidden potential in your data!

Our game-changing free guide, "How to Use AI for Data Analysis", is your key to unlocking a world of insights you never knew existed.

Discover how to:

Uncover patterns and trends invisible to the naked eye
Gain a cutting-edge skill that sets you apart in the data-driven world
Turn overwhelming data into clear, actionable insights
Master AI techniques without getting lost in technical jargon

Whether you're a curious beginner or a seasoned analyst, this guide opens doors to a new realm of data mastery.

Download Now!

Anthropic’s new Claude prompt caching will save developers a fortune

Anthropic has introduced a new feature called prompt caching on its API, which is available in public beta for Claude 3.5 Sonnet and Claude 3 Haiku models, with support for Claude 3 Opus coming soon. This feature allows developers to store frequently used contexts between API calls, reducing the need to resend large prompts and resulting in significant cost savings. Early users have reported substantial speed and cost improvements in various use cases, including embedding entire documents, providing long instructions, and autocompleting code.

Prompt caching offers a pricing advantage by reducing the cost per token. For example, in Claude 3.5 Sonnet, writing a prompt to be cached costs $3.75 per million tokens, but reusing that cached prompt costs just $0.30 per million tokens. Similarly, Claude 3 Haiku users will pay $0.30 per million tokens to cache and $0.03 per million tokens to access stored prompts. While caching is not yet available for Claude 3 Opus, Anthropic has published its prices, showing a significant cost reduction when using cached prompts.

Prompt caching differs from large language model memory, as it stores and recalls actual prompts and responses rather than just remembering preferences or details. While other platforms like Lamina offer similar caching mechanisms to reduce GPU costs, Anthropic's implementation focuses on optimizing costs for developers building conversational agents and other tools. However, it’s important to note that the cached prompts have a 5-minute lifetime, which refreshes with each use. This feature positions Anthropic in a competitive landscape, as AI platforms race to offer low-priced options for developers.

The Google Pixel 9’s AI Camera Features Let You Reshape Reality

Google's Pixel 9 series takes mobile photography to new heights by integrating advanced AI features that reshape photos with ease. Key innovations include Add Me, which allows users to superimpose multiple people into one shot without needing a third photographer, and Reimagine, which uses AI to alter scenes or add elements like stormy clouds with just a text prompt. These features expand beyond traditional photo editing and make it accessible to anyone, regardless of technical expertise.

Other tools, like Autoframe and Zoom Enhance, help fine-tune images post-capture. Autoframe adjusts composition by generating pixels to reposition subjects, while Zoom Enhance sharpens zoomed-in shots using AI to fill in missing details. These features are part of the Pixel 9's enhanced HDR+ pipeline, offering users more control over their photos without requiring manual adjustments in editing software.

The Pixel 9 series also improves video capture through Video Boost, which processes clips in the cloud for superior quality, especially in low-light conditions. Additionally, the Panorama mode has been rebuilt with better HDR+ and Night Sight capabilities for clearer, more detailed wide shots. Altogether, the Pixel 9's camera enhancements reflect Google's continued push to blend AI and photography for both novice and professional users.