Anthropic to share Mythos cyber flaw findings with global finance watchdog
Startup has declined to release Claude Mythos AI model publicly amid fears it could be used by hackers
TL;DR
- Anthropic will brief the Financial Stability Board (FSB) on its Claude Mythos AI model.
- The AI model has advanced capabilities in highlighting IT system flaws that could be used by hackers.
- Mythos has not been released publicly but is being tested by tech companies and banks like Apple and JP Morgan.
- The UK's AI Security Institute (AISI) noted a 'notable capability jump' in Mythos, which successfully completed a cybersecurity test.
- The FSB monitors and makes recommendations on the global financial system.
- The International Monetary Fund (IMF) has also highlighted rising financial stability risks due to AI developments.