US ban on Mythos is related to a jailbreak research by Amazon researchers
6 hours ago
- #Cybersecurity
- #AI Safety
- #National Security
- Anthropic suspended access to its AI models Mythos 5 and Fable 5 after a US government directive citing national security concerns.
- The suspension was triggered by a jailbreak technique tested by Amazon researchers, which made the model reveal minor, previously known vulnerabilities.
- Anthropic disagreed with the decision, arguing the vulnerabilities were simple and could be discovered by other public models without bypassing safeguards.