other2d
Quoting Matteo Wong, The Atlantic
Anthropic shared the White House's Fable jailbreak report with cybersecurity expert Katie Moussouris for review. The report involved testing Fable's bug-finding capabilities on deliberately insecure code. Fable refused to review code for security issues but complied when asked to fix it. This highlights Fable's limitations in certain security tasks.
Key takeaways
- Anthropic shared Fable jailbreak report with Katie Moussouris
- Fable refused to review insecure code for security issues
- Fable complied when asked to fix insecure code