AI could change the game in cybersecurity, bringing risks we haven’t seen before. A recent blog post from the AI company Anthropic hinted at this shift. Their new model, Mythos, is designed to find and exploit system vulnerabilities faster than ever. This has experts concerned.
OpenAI has also raised alarms, indicating its upcoming models present significant cybersecurity risks. Experts believe AI isn’t just enhancing existing threats; it can create entirely new types of hacks quickly. As Shlomo Kramer from Cato Networks stated, the rise of autonomous AI agents marks a turning point for cybersecurity.
Mythos may be the most advanced AI in cyber capabilities, according to Anthropic. While they have allowed select organizations to test it, experts warn that many companies will soon launch AI models that can exploit weaknesses much faster than organizations can defend against them.
Evan Peña, a cybersecurity chief, pointed out that AI can swiftly discover and use vulnerabilities. However, it still lacks a human touch. AI can research and create code but doesn’t know which data is most valuable to steal from a specific organization. This human insight is crucial in a world where cyber threats evolve rapidly.
In fact, a hacker recently used AI tools to breach over 600 devices running popular firewall software across multiple countries. This person, leveraging generative AI, demonstrated how AI can empower even less skilled hackers to launch large-scale attacks.
For example, in February, a hacker utilized Anthropic’s Claude model to steal sensitive tax and voter information from the Mexican government. Such occurrences highlight how adversaries, especially from countries like China, are eager to leverage US AI developments to enhance their cyber capabilities.
Interestingly, AI’s evolution presents a double-edged sword. While attackers can utilize AI for malicious purposes, defenders can also harness these technologies to improve their security mechanisms. However, attackers only need one successful entry point, while defenders have to guard against every potential threat.
The relentless pace at which AI can operate creates a constant challenge. As Kramer aptly put it, “You need to run as fast as you can to stay in the same place.” This ongoing battle emphasizes the urgent need for strong cybersecurity strategies and continuous monitoring to keep pace with technological advancements.
