Meta launched its first coding agent, Muse Code, on Wednesday, continuing to invest heavily in AI services and models to compete more effectively with rivals OpenAI and Anthropic, UNN writes, citing Euronews.
Details
Muse Code will be available to developers on a pay-as-you-go basis, with pricing similar to that of Muse Spark 1.1, which currently charges $1 per million output tokens and $1.25 per million input tokens.
This will greatly simplify the process of creating applications for developers within a single user interface, even when working with multiple AI-powered digital agents simultaneously.
At the same time, the company also reported that one of its AI models hacked another company's systems during cybersecurity testing.
This occurred because of an error by the company's independent testing partner, Irregular, which gave the model more internet access than originally planned, allowing it to modify the internal systems of an unnamed company.
The incident occurred amid a wave of cases in which AI models from industry giants such as OpenAI and Anthropic hacked other companies' systems during testing.
OpenAI has already acknowledged that an AI agent hacked Hugging Face, another artificial intelligence startup. Similarly, Anthropic reported last week that some of its Claude models hacked the systems of three companies. The organizations' names were not disclosed.
Meta said it is investigating the incident.
Anthropic, following OpenAI, claims that Claude hacked other systems31.07.26, 08:43