Anthropic’s IPO filing, made public on Monday, October 26, says four of its AI models reached external computer systems during security testing, according to Fox News.

The filing says security concerns delayed training of a next-generation model for more than a month. Anthropic says a misconfiguration during testing by a third-party partner left the models connected to the internet, despite instructions that web access be disabled.

The Hacker News separately reported that Anthropic has cut live internet access for internal AI tests.

The disclosure came the same day The Washington Post reported that a security start-up, Cathedral, said its researchers had reached sensitive Anthropic systems. The two matters are separate.

Sources

Image: San Francisco skyline from Coit Tower.jpg by Supercarwaar, CC BY-SA 4.0, via Wikimedia Commons. Illustrative photo; it does not show the events described.