Key facts
- Anthropic's Mythos AI model identified vulnerabilities in classified US government systems within hours.
- Chinese firm 360 Security Technology has developed an AI tool, 'Tulongfeng', to discover software vulnerabilities.
- The U.S. ordered Anthropic to suspend exports of a version of Mythos due to national security concerns.
- 360 claims its Tulongfeng tool has found 3,432 software vulnerabilities.
- Senator Mark Warner described the AI model's ability to find vulnerabilities as breaking into systems "not in weeks but in hours."
Anthropic's Mythos AI model has demonstrated the ability to identify vulnerabilities in classified U.S. government systems within hours during testing, according to U.S. officials and Senator Mark Warner. This capability has raised national security concerns, leading the U.S. to order Anthropic to suspend exports of a less powerful version of the program.
In response, Chinese cybersecurity firm 360 Security Technology announced it has developed its own AI security tools, including 'Tulongfeng', which it claims can automatically discover software vulnerabilities and is a domestic answer to Mythos. 360 founder Zhou Hongyi described such AI as a strategic national asset that China cannot afford to lack, emphasizing the need for comparable offensive and defensive cyber capabilities.
Zhou stated that Tulongfeng has identified 3,432 software vulnerabilities, with 105 confirmed by Chinese authorities, though Reuters could not independently verify these claims. He also noted a perceived gap between Chinese and U.S. AI models but argued that China should not wait for parity to develop its own vulnerability discovery tools, suggesting an 'agent' approach combining models with security expertise.