Anthropic Releases Claude Fable 5, But With Caveats

A highly anticipated AI model has finally been made available to the public. However, it's not quite what many were expecting - and users will need to be cautious when using it.

Claude Fable 5 is a powerful tool that Anthropic claims could have 'profound good' for the world. But its extreme cybersecurity capabilities also pose a risk of being misused to cause significant damage if left unchecked.

The company has implemented conservative safeguards in an effort to mitigate this risk, which may sometimes result in harmless requests being denied - although Anthropic estimates this will occur less than 5% of the time.

A separate version of the model, Claude Mythos 5, is also being released for a limited group of cyberdefenders and infrastructure providers. This variant has some safeguards lifted, but it's not clear what specific changes have been made.

Benchmarks suggest that Fable 5 offers significant improvements over previous models in areas such as coding and spatial reasoning - although the biggest gains are seen in cybersecurity.

For paid Claude users, there is both good news and bad. On one hand, subscribers can access Fable 5 immediately, including those on plans like Pro, Max, Team, and Enterprise. However, Anthropic will be removing Fable 5 from these plans on June 23rd, requiring users to purchase pre-paid usage credits instead.

The company has stated its intention to restore Fable 5 as a standard part of subscription plans when sufficient capacity allows it - but no specific timeline has been given for this.