Israel’s cybersecurity industry, the most prominent exporter in the country’s tech sector, went on high alert last weekend following the launch of Anthropic’s new AI tool Claude Code a multi-task cybersecurity system that detects data security vulnerabilities in code. The biggest cybersecurity stocks fell sharply on Wall Street, including Crowdstrike, Palo Alto Networks, Cloudflare and more, and it is clear that attention will be focused on Israeli cybersecurity enterprises as well. "The absurdity is that these cybersecurity companies are not even actually harmed by the tool, because they are not engaged in detecting code weaknesses, but in securing endpoints, identities, or cloud systems - areas that Claude cannot easily replace because it requires intimate knowledge of the client organization," says the CEO of an Israeli cybersecurity company, who preferred not to be identified. "This indicates the lack of knowledge of investors in the capital market more than the companies themselves."
"The tool primarily serves Anthropic"
Israeli cybersecurity company Tenzai CEO Pavel Gurvich says, "Anthropic did not develop this tool simply to compete in the industry, but primarily to solve an internal problem. It is interested in launching code development tools that are as bug-free and free of vulnerabilities as possible, so this product primarily serves it. But it is related to Anthropic as a provider of an engine writing code and not touching many other systems besides code that are in the organization."
Nevertheless, several companies may be directly affected by the launch of the tool, veteran companies whose products scan code, such as Israeli companies Snyk and Checkmarx and US company GitHub, with its Github Advanced Security product).
While these companies are making efforts to adapt themselves to the AI era, investors prefer to put their money in newer companies that were founded in the era in which AI agents develop applications themselves, while creating many security vulnerabilities. In response to the rise of agents who write code and discover vulnerabilities themselves, these companies have launched their own solutions: Snyk launched Evo, and Checkmarx launched Developer Assist.
"The Developer Assist family of solutions was created with the aim of securing the entire supply chain of agent-based software development, from initiation, writing, and implementation stages," said Checkmarx, which added that the company was prepared for this development by Anthropic "in advance."
The companies that don't feel threatened
It's too early to tell whether in this arms race between software companies and AI giants, Anthropic or OpenAI has the upper hand.
Ostensibly, Claude may also have a short-term impact on the market for cybersecurity companies that focus on protecting software supply chains - that is, on code repositories of applications that are connected to each other. Companies that have raised tens of millions of dollars each are active in this field in Israel, such as Cycode and Legit Security.
However, each of these Israeli companies boasts a relative advantage that Claude does not have. While the tool can detect vulnerabilities in the code, it has difficulty detecting a hack in retrospect or a leak of secrets, or prioritizing different databases. In other words, while Claude functions as a "code scanner" that can sometimes also fix code vulnerabilities, it lacks recognition of the organization's structure and the sensitivities involved.
Cycode CEO Lior Levy says, "It's a spot-on scanning tool, one small component of the world of application risks. Organizations are not just looking to identify vulnerabilities in the code, but to manage risk holistically. As scanning tools improve, the value of a platform that centralizes, prioritizes, and manages organizational risk increases. From our perspective, this is a tool that we can integrate with and enrich the data from." As a result, Israeli companies active in the penetration testing market, remain unruffled.
Gurvich, who develops intelligent AI agents at Tanzei, which perform penetration tests, does not feel threatened: "Many times the vulnerabilities are not in the code at all, and sometimes you need to run active tests on infrastructure and organizational applications. To do this well, you need to collect relevant data, test performance in labs of different infrastructures and cloud providers, and also on the company's local website."
Another industry source also warns against the new tool. "In our world, it is almost unreasonable to let AI tools like Claude 'loose'; such tools can plan an attack that will cause the collapse of the organizational defense system while it is working."
Congratulations to those who bought cybersecurity stocks at a discount
Meanwhile, on the ground, the major companies in the field in Israel are adapting themselves to the new era. Check Point rushed to acquire three companies that are introducing smart agents into its long-standing product, while Zafran, which deals with detecting cybersecurity vulnerabilities, has developed smart agents that can identify and close loopholes.
Zafran Security VP marketing Nick Fisher reassured his online followers over the weekend and claimed that Anthropic does not want to compete with cybersecurity companies, but rather to become the computing layer on which they are built. "Amazon also competed for years in identifying cybersecurity threats in the cloud, but it made the real money in the cloud service. Anthropic is playing the same game. Congratulations to everyone who bought the cyber stocks at a discount last Friday."
Glilot Capital Partners managing partner Kobi Samborsky says, "There is a perfect storm here. On the one hand, there is an opportunity to start a lot of new companies, and on the other hand, there is a real threat to existing companies. The change is so dramatic that it creates both. But there is also a degree of naivety in claiming that a general machine will solve everything, and there will be no need for vendors."
He adds,, "AI engines rely on knowledge and data, and some of this data is with the cybersecurity players. And besides, cybersecurity companies don't just sell technology, they sell responsibility and trust. When you buy a product from Palo Alto, you transfer some of the responsibility to the vendor. When you build a solution with AI yourself, the responsibility remains with you."
The CEO of an Israeli cybersecurity company concludes, "Even if Israeli cybersecurity companies are not harmed by Claude in the first stage, the mere fact that their investors know that there are AI engines that can do some of their work will reduce their ability to hire the number of employees they wanted and the amounts of money and valuations desired in future financing rounds. The cybersecurity industry is already changing, and everyone will have to adapt to it."
Published by Globes, Israel business news - en.globes.co.il - on February 24, 2026.
© Copyright of Globes Publisher Itonut (1983) Ltd., 2026.