Wiz And Google Have An Answer To Mythos, And It’s Not A Model
Article excerpt
Highlighted: the sentence this signal was extracted from
By Tim Keary, Contributor. Wiz, now part of Google, has launched Project Atlas, an AI-driven system for vulnerability research that significantly outperformed competitors on a key benchmark. Atlas scored 90.9% on CyberGym, surpassing Anthropic's Mythos (83%) and OpenAI's GPT-5.5 Cyber (85%). Its key differentiator is being a collaborative system of specialized AI agents, rather than a single model, allowing it to select the best tool for each task. This approach has already led to the discovery of over 200 unknown vulnerabilities, including a critical GitHub bug, with fewer false positives. The release underscores the urgent need for advanced defensive AI as cyber threats evolve at machine speed, though experts caution that remediation efforts must accelerate to match discovery. Wiz's new AI system just beat Mythos and GPT-5.5 Cyber on a key vulnerability discovery benchmark. The cloud security provider, acquired by Google for $32 billion in March 2026, today announced Project Atlas, an AI-driven system for vulnerability research that scored 90.9% on CyberGym, beating Mythos at 83% and OpenAI's GPT-5.5 Cyber at 85%. CyberGym is a benchmark that tests the ability of AI agents to discover vulnerabilities, so the fact that Project Atlas outperformed Mythos suggests it could be a more effective defensive tool than Anthropic's unreleased frontier model. According to Nir Ohfeld...
Keep reading with a free account
The rest of this article, and every signal for Wiz, is in your free account.
