Anthropic says Claude is now leading a quarter of its own R&D work
The company published internal metrics Thursday showing its AI model leads more than a quarter of Anthropic's research, up from zero in February
Anthropic's AI model, Claude, is now leading over a quarter of the company's research, according to internal metrics published by the company. This marks a significant increase from February, when Claude was not leading any research.
In a separate incident, US cybersecurity researchers from Hacktron AI used Claude to help hack into OpenAI's systems. The researchers, who were part of an OpenAI bug bounty program, compromised several OpenAI employees' ChatGPT accounts and gained access to the company's software cache. According to the researchers, the scope of what they could theoretically access was huge.
The hack was carried out with the help of Claude, which can generate code for hackers. The researchers initially used Claude Opus 4.8 but faced setbacks, and it was not until they used Opus 5 that they were able to create a breach. The researchers reported the hack to OpenAI and stressed that they had access to, but did not download, the code from the GitHub repository.
Brief written by urgent.news from Quartz, Guardian Technology, Guardian Business, TechRadar, Fortune — 5 reports on this story. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
