Anthropic's AI will now tell users when requests are downgraded for national security after backlash
Fortune • • business
Key Points:
- Anthropic released Fable 5, a Mythos-class AI model with advanced capabilities, after initially withholding such models due to safety concerns about cybersecurity risks.
- Fable 5 includes hidden safeguards that silently downgrade requests related to advanced AI development, prompting criticism from researchers who argue this slows AI progress.
- In response to backlash, Anthropic announced it will increase transparency by visibly indicating when requests are downgraded or refused, while maintaining restrictions to prevent use in creating competing AI systems.
- The company cited national security concerns as a reason for some restrictions, aiming to prevent foreign adversaries from enhancing their AI capabilities at the expense of the U.S.
- Anthropic’s safety measures and government tensions underscore the growing intersection of AI development, safety, and national security, especially amid ongoing disputes with the Department of War and its recent confidential IPO filing.