Codenamed “Mythos” under the broader “Capybara” testing tier, this new AI possesses advanced capabilities for discovering complex software vulnerabilities, triggering a sharp sell-off across the security sector.
The financial market reacted quickly, driven by fears that autonomous AI agents will disrupt traditional enterprise security architectures and outpace human defenders.
Internal reports reveal that Mythos represents a massive leap in technical performance.
According to Anthropic’s leaked draft blog, the Capybara tier is described as “a new name for a new tier of model: larger and more intelligent than our Opus models, which were, until now, our most powerful.”
It significantly outperforms Anthropic’s previously leading model, Claude Opus 4.6, in software coding, academic reasoning, and cybersecurity benchmarks.
An Anthropic spokesperson confirmed to Fortune that the new model represented “a step change” in AI capabilities, calling it the “most capable we’ve built to date.”
The most disruptive feature for the security industry is the model’s ability to autonomously hunt down previously unknown flaws, including potential zero-day vulnerabilities, within live production code.
Anthropic is currently restricting the model to a small group of highly vetted early-access customers, recognizing the unprecedented nature of this technology and deliberately pacing the release.
Unlike older security tools that rely on static, rule-based scanning, Mythos leverages dynamic reasoning.
It analyzes complex codebases much like a human security researcher would, tracing intricate data flows to uncover hidden weaknesses.
This capability was already foreshadowed in February 2026, when Anthropic’s Claude Code Security tool demonstrated that its Opus 4.6 model had identified over 500 vulnerabilities in production open-source codebases during internal testing, bugs that had survived decades of expert review.
The anxiety surrounding this new technology triggered a rapid sell-off across the cybersecurity sector.
The Global X Cybersecurity ETF dropped by 4.5% on Friday, marking its lowest close since November 2023.
This recent slide pushes the fund’s year-to-date decline past 21%, reflecting growing confidence in traditional security vendors.
Bernstein analyst Peter Weed noted that investors started selling “like clockwork,” referencing the now-familiar pattern of cybersecurity stocks falling every time Anthropic announces a significant product development.
However, Weed also cautioned that the selloff may be overdone, writing that the news “does not appear to reduce the potential cybersecurity sector tailwinds” from generative AI and large language models.
The core issue shaking the industry is the dual-use nature of the Mythos model.
Anthropic explicitly warns in its internal assessments that Mythos is “currently far ahead of any other AI model in cyber capabilities” and that it “presages an upcoming wave of models that can exploit vulnerabilities in ways that far outpace the efforts of defenders.”
This power raises serious alarms about an impending wave of offensive AI tools that could exploit vulnerabilities much faster than human developers can patch them.
These fears are grounded in the recent historical context. This latest market shock compounds friction that began in February 2026 with the launch of Claude Code Security Anthropic’s AI-powered tool that autonomously finds and repairs software vulnerabilities.
Furthermore, well-resourced threat actors are already attempting to weaponize these advanced models.
Anthropic previously disclosed that a state-sponsored hacking group from China had already attempted to use earlier versions of Claude to automate attack sequences, infiltrating numerous organizations in a documented real-world test of the technology’s offensive potential.
Anthropic’s rollout strategy attempts to thread this needle. The company is offering early access to the Capybara tier specifically to cyber defenders, aiming to give them a “head start in improving the robustness of their codebases against the impending wave of AI-driven exploits.”
Follow us on Google News , LinkedIn and X to Get More Instant Updates. Set Cyberpress as a Preferred Source in Google
The post Cybersecurity Stocks Slide as Anthropic Tests Advanced AI Model appeared first on Cyber Security News.
A Russian state-sponsored hacking group known as Sandworm has been caught making a calculated pivot…
A Chinese state-linked hacking group known as FamousSparrow has quietly infiltrated an Azerbaijani oil and…
INDIANAPOLIS, Ind. (WOWO) — Richard Allen’s attorneys will get their day in court as they…
MUNCIE, Ind. (WOWO) — A man wanted in connection with a Muncie shooting was apprehended…
Protestors at the Tennessee Capitol pushed back against a redistricting plan that Republicans ultimately enacted.…
The design of Xbox's upcoming cloud gaming controller has leaked online and it looks pretty…
This website uses cookies.