Satya Nadella criticizes Anthropic blocking Claude fable outputs for training
Microsoft's Satya Nadella isn't happy with how Anthropic restricts its latest AI, Claude Fable.
He called the rules something that "don't make sense" because they block users from using Claude's outputs to train other competing models.
Researchers and users have also noticed that when asking tricky questions, especially about cybersecurity or advanced AI, Claude Fable sometimes switches to its older version, Claude Opus 4.8.
Anthropic defends switching to older models
Anthropic says these limits are there to prevent misuse, especially on risky topics.
Its system reviews conversations and files, and will swap in an older model if things get too sensitive.
You can turn off this automatic switch, but you might need to tweak your prompts or pick a different model yourself.
All this is happening as the AI industry heats up, with companies wanting flexible and affordable tools that don't compromise on safety.