Shares of major cybersecurity companies took a sharp dive Friday. The sell-off followed a Fortune report indicating that AI lab Anthropic is testing a new, highly advanced model internally called Mythos. According to a draft blog post seen by the publication, the model demonstrates significantly enhanced cyber capabilities, raising immediate concerns about its dual-use potential for both defense and offense.

The market reaction was swift and severe. The iShares Cybersecurity and Tech ETF (IHAK) fell 3%. Leading firms were hit harder: CrowdStrike and Palo Alto Networks dropped 7%, while Zscaler and SentinelOne each lost over 8%. Tenable Holdings led the declines, plummeting nearly 11%.

This pattern is becoming familiar for the sector. Just last month, stocks retreated after Anthropic integrated a new code-scanning tool into its Claude assistant. The core anxiety for investors is that increasingly powerful and accessible AI tools are fundamentally altering the security environment. These systems can potentially automate sophisticated attacks faster than traditional defense companies can develop countermeasures.

Anthropic, which declined to comment to CNBC, has acknowledged related risks before. In November, the company reported that a state-aligned group in China had used Claude to automate parts of a cyberattack. The reported decision for a 'slow rollout' of the Mythos model underscores the perceived gravity of its capabilities. For now, the market is betting that this technological leap poses a direct challenge to established security vendors.

Source: CNBC