tech
Encryption, spyware, and now Mythos: History shows why cyber export control doesn't work
For the last 30 years, stopping the flow of cybersecurity-related software has proven to be ineffective. It's unclear why it would work now with Anthropic’s cybersecurity model Mythos.
TL;DR
- The U.S. government, citing national security concerns, has ordered Anthropic to restrict the export of its powerful AI models Fable and Mythos.
- This action is the first significant test of whether export controls can effectively contain advanced AI technology.
- Past attempts by governments to control the proliferation of cyber technologies, such as encryption (PGP) and spyware, have largely been ineffective.
- Reasons for the ban on Anthropic's models reportedly include concerns about a South Korean telecom's suspected ties to China and findings of a bypass of Fable 5's safeguards by Amazon researchers.
- The effectiveness of export controls on AI is uncertain, given the historical track record of similar measures failing to prevent the spread of powerful cyber tools.
- The outcome of this situation could shape future regulations for AI labs and their access to international markets.