GitHub Spark lets you build web apps in plain English

GitHub has unveiled "Spark," a new experimental tool designed to simplify web app creation using natural language. Announced at the GitHub Universe conference, Spark is part of GitHub Next Labs, aimed at allowing users to build web apps without any coding knowledge. It generates apps based on user prompts and provides a live preview almost instantly. While experienced developers can still view and adjust the underlying code, Spark is built to let users interactively create prototypes and fine-tune them through a conversational interface.

According to GitHub CEO Thomas Dohmke, Spark is intended for building “micro apps” — small, practical tools for everyday tasks. However, GitHub has not imposed strict limitations on the complexity of these apps, and the tool’s capacity remains unrestricted during its technical preview. Spark supports various web APIs, enabling users to build increasingly sophisticated applications. This flexibility also lets users test new ideas or experiment with software development, relying solely on natural language inputs.

The tool offers users the option to choose between Anthropic’s Claude Sonnet and OpenAI’s GPT models, reflecting GitHub’s broader Copilot functionality, which also allows model selection. Apps built with Spark can be shared with customized access controls, and the shared code is fully accessible, allowing other users to enhance or build upon it. For those with coding knowledge, Spark offers the additional benefit of manual editing, which is useful when refining or correcting AI-generated code.

In essence, Spark’s introduction highlights GitHub’s ongoing commitment to democratizing software creation. By making app development more accessible, GitHub aims to empower users to create digital solutions for day-to-day tasks and explore new possibilities in software through natural language, sparking creativity without needing in-depth programming knowledge.

Finally, a developer tool that makes digging around for the answer a thing of the past.

Other developer tools can’t tell you how your codebase works and why. Unblocked can. We augment your source code with context from Slack, Confluence, Jira (and more), so your team gets helpful answers instantly.

Unblocked gives you:

• Instant answers about every aspect of your codebase 

• Relevant, historical context for any file open in an IDE

• A powerful Slack integration that answers questions in channels, removing interruptions to you

⌛Engineers who use Unblocked save an hour or more a day.

“Unblocked empowers every developer with the insight once reserved for the seasoned few. ” Mathew Spolin - SVP Engineering at AppDirect

A mysterious new image generation model has appeared

A new image generation model, known as “red_panda” has recently emerged as a top performer, surpassing popular models from Midjourney, Black Forest Labs, and OpenAI on the crowdsourced benchmark, Artificial Analysis. This model has shown a substantial lead, sitting about 40 Elo points above Black Forest Labs’ Flux1.1 Pro, the second-highest-ranking model on the text-to-image leaderboard. The Artificial Analysis leaderboard utilizes an Elo ranking system, commonly associated with assessing chess player performance, to determine the relative strength of image generation models.

Artificial Analysis functions similarly to the community-driven benchmark Chatbot Arena by ranking models through crowdsourced voting. Users are presented with images generated by two randomly selected models based on the same prompt and are asked to vote on which one best reflects the prompt’s intent. While this process can introduce bias, as most voters tend to be AI enthusiasts, it still provides valuable insights into model performance based on user preference.

Apart from its ranking, red_panda is notable for its generation speed. It reportedly takes only around 7 seconds to generate an image, making it roughly twice as fast as OpenAI’s DALL-E 3, which gives it a significant advantage in scenarios requiring rapid output.

UK man who used AI to create child sexual abuse imagery sentenced to 18 years in prison

Hugh Nelson, a 27-year-old from Bolton, has been sentenced to 18 years in prison for using AI to create explicit child abuse images from ordinary photos. This case marks a legal milestone in the UK as the first conviction involving AI-generated abuse imagery. Nelson manipulated images of real children into disturbing, exploitative content using Daz 3D, an AI-powered software. Some images were made at the request of paedophiles, who provided him with pictures of children they knew. His illicit activity was uncovered when he revealed his pricing to an undercover officer, and upon arrest, Nelson confessed his actions had spiraled out of control.

Greater Manchester Police (GMP), who led the investigation, emphasized the growing difficulty in policing digitally manipulated images, as they present new challenges to current child protection laws. Detectives uncovered that Nelson not only sold the images online, earning approximately £5,000, but also encouraged others to commit abuse, communicating with individuals worldwide. Sentencing judge Martin Walsh described Nelson’s content as "harrowing and sickening," with no regard for the harm caused by his actions.

This case has prompted law enforcement to consider the rapid evolution of AI technology, which, although useful in investigations, is also exploited for harmful purposes. DCI Jen Tattersall of GMP noted that computer-generated child abuse imagery is becoming more common, requiring police forces to adapt. The conviction underscores the CPS's commitment to prosecuting such crimes, sending a firm warning to those misusing AI to produce exploitative content.

Other stuff

• GitHub Copilot will support models from Anthropic, Google, and OpenAI

• GitHub’s Copilot is also coming to Apple’s Xcode

• Google CEO says more than a quarter of the company's new code is created by AI

• OpenAI reportedly planning to build its first AI chip in 2026

• LinkedIn launches its first AI agent to take on the role of job recruiters

• Google says its next-gen AI agents won’t launch until 2025 at the earliest

• Tony Fadell takes a shot at Sam Altman in an interview

• A look inside the data center behind Elon Musk's Grok AI

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

Hope you enjoyed today's newsletter

⚡️ Join over 200,000 people using the Superpower ChatGPT extension on Chrome and Firefox.

OR