Our readingWhat is a korrent?
At Anthropic, a less severe version of an OpenAI-style rogue-AI incident is known to have happened, and worse undisclosed internal incidents seem likely.
Drawn from what Zvi Mowshowitz said
What this subject means
cybersecurity Attack and defence in software: vulnerabilities, hardening, and what changes when a model is good at finding them.
In the views onAnthropicunder our view: The lab's safety record is asserted from the inside and doubted from the outside