Cybersecurity stocks sink on report of Anthropic's new model
Software stocks plunged last month after Anthropic released a new AI-powered plugin for Claude aimed at the legal industry that put pressure on shares of software as a service (SaaS) companies, creating a loss of $300 billion in market value in a single day.
The Claude release created a doomsday scenario among investors who grew concerned about the software industry facing extreme disruption from artificial intelligence. The bloodbath among software stocks got so bad that it earned the nickname "SaaSpocalypse" among investors on social media.
And while software stocks have still yet to fully recover, Anthropic struck again last week after Fortune got access to a draft of a blog post detailing the company's new AI model that had been inadvertently stored in a publicly accessible data cache.
After Fortune reported on the details of Anthropic's new AI model, called Claude Mythos, it sent cybersecurity stocks tumbling on Friday. But unlike with what happened to SaaS stocks in February, the latest selloff wasn't driven by Claude offering a new model that could compete with the services that the cybersecurity firms provide.
Instead, it's the threat that Claude's new system will be too powerful for cybersecurity firms to protect against.
Anthropic said in the draft of the blog post that Mythos is “by far the most powerful AI model we’ve ever developed." In the same draft, the company also refers to a new tier of AI models called Claude Capybara, which it calls "larger and more intelligent than our Opus models—which were, until now, our most powerful.”
According to Fortune, Mythos and Capybara appeared to be referring to the same underlying model.
But what sent cybersecurity stocks down was that Anthropic claims in its draft that Capybara is “currently far ahead of any other AI model in cyber capabilities.”
The company essentially fears that hackers could use the system to wage attacks that cybersecurity firms would be incapable of preventing. While cybersecurity firms like CrowdStrike (CRWD) have been pouring money into building out their AI capabilities, Anthropic said that Capybara "presages an upcoming wave of models that can exploit vulnerabilities in ways that far outpace the efforts of defenders."
Both CrowdStrike and Palo Alto Networks, Inc. (PANW) fell nearly 6% on Friday. Tenable Holdings, Inc. (TENB) plunged 9.6%, SentienelOne, Inc. (S) slid over 6% and Okta, INc. (OKTA) sank more 7.8%.
While a number of Wall Street insiders have been pushing back on the doom narrative facing the software industry, seeing many software companies using AI to expand their businesses rather than be disrupted by the new technology, it remains to be seen whether this latest selloff will create a similar push back.
After all, there are more unknowns than knowns about what cybersecurity firms are facing with Anthropic's new model.
In fact, Anthropic is currently giving early access to companies in the cybersecurity industry, with Anthropic noting in its blog post that it is "giving them a head start in improving the robustness of their codebases against the impending wave of AI-driven exploits.”
“In preparing to release Claude Capybara, we want to act with extra caution and understand the risks it poses—even beyond what we learn in our own testing," the company said in its draft. "In particular, we want to understand the model’s potential near-term risks in the realm of cybersecurity—and share the results to help cyber defenders prepare."