Meta entered the race for automated coding tools on Wednesday when it announced a new app for developers of artificial intelligence products, as it competes with other major AI labs for customers and revenue.

Dubbed Muse Code, Meta said in a blog post that the new product is an agentic coding tool that can perform "complex software engineering tasks" with large amounts of data, including planning, writing and validating code.

Agentic tools are software applications which are programmed to handle tasks without human supervision, though often someone will still make the final call on decisions or supervise the execution of tasks.

Muse Code launched in beta, or in a preview mode.

Meta also released an updated version of its flagship model, known as Spark, which it unveiled in April.

The update, called Spark 1.2, is "coding-focused" and available through Muse Code and another developer application called Meta Model API, the social media giant said.

Meta has poured billions into its AI division, now known as Meta Superintelligence Labs, as it attempts to keep up with OpenAI and Anthropic.

Major AI developers are in a tight race to dominate the sector, not just in pure technical capabilities but also to prove that their heavy spending to meet growing demand will turn into profits.

Amazon, Microsoft, Alphabet and Meta are collectively on track to pour around $700 billion into AI data centers, chips and computing infrastructure this year.

While chatbots have generated a frenzy around AI since OpenAI launched ChatGPT in late 2022, few people have been willing to pay for them. However, software engineers and companies are increasingly willing to pay for automated coding tools.

In February, OpenAI said ChatGPT reached 900 million weekly active users, including 9 million businesses and 50 million consumers that pay for it.

Anthropic has not disclosed its total users, but the company told TechCrunch in March that its paid subscriptions "have more than doubled this year."

SpaceXAI, a subdivision of Elon Musk's rocket company SpaceX, develops its own chatbot called Grok, and the company announced earlier this year it would acquire coding startup Cursor for $60 billion.

Microsoft, which has invested billions into both OpenAI and Anthropic, offers coding tools through its business-focused AI app, Copilot, in addition to a developer product, GitHub.

Recent cyberattacks carried out autonomously by advanced AI models during routine testing, like those developed by OpenAI and Anthropic, have also raised concerns about their capabilities.

A model developed by Meta, Spark 1.1, was involved in a similar incident, The Information reported on Wednesday.

That incident was caused by a "misconfiguration" at a third-party testing partner, a Meta spokesperson told AFP in a statement.

"We are currently investigating and will issue a full retrospective once we have all the facts," the statement said.