Anthropic Debuts Security Tool as Cybersecurity Stocks Fall - PYMNTS.com

Anthropic Debuts Security Tool as Cybersecurity Stocks Fall BY PYMNTS  |  FEBRUARY 22, 2026  |  Cybersecurity software stocks fell last week after Anthropic unveiled a new security feature for its Claude model. That’s according to a report Friday (Feb. 20) from Bloomberg News, which characterizes this as the latest example of software shares dropping due to worries about competition from artificial intelligence firms. In this case, the release of Anthropic’s tool, which “scans codebases for security vulnerabilities and suggests targeted software patches for human review,” prefigured a decline in the price of companies such as Cloudflare, the report said. “There’s been steady selling in software, and today it’s security that’s getting a mini-flash crash on a headline,” said Dennis Dick, head trader at Triple D Trading.  “This kind of market is scary for investors, because things are just moving relentlessly to the downside as soon as you get a hint of disruption. It’s rational to be cautious, because people were saying a while ago that the software drop was overdone, and yet it keeps going down.” Another AI-related software selloff occurred last summer, affecting the likes of Salesforce and Workday, PYMNTS reported. We’d love to be your preferred source for news. Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks! Add as Preferred Source Advertisement: Scroll to Continue The Bloomberg report added that much of the selling has accompanied the launch of new AI tools from the likes of Anthropic, Google and OpenAI. Investors worry that the ability to “vibe code,” or employ AI to write software code, will let users develop their own applications, meaning less demand for legacy software products. However, as PYMNTS wrote last year, research has shown that vibe coding is not about to replace human software developers anytime sooner. According to that report, researchers have found that agentic AI models such as Claude performed best when developers reviewed outputs after key checkpoints instead of running fully autonomous sessions.  “Without those checkpoints, the models produced longer, less maintainable codebases and missed security constraints,” PYMNTS added. The findings line up with earlier research on CoAct-1: Computer-Using Agents with Coding as Actions, which also concluded that human interaction remains critical for steering multi-agent software systems toward reliable outcomes. “Vibe coding may indeed spark a new economy, but not through total automation. Its real promise lies in redefining collaboration: Developers who manage, teach and correct AI will shape the next era of software creation,” that report said. “In the process, coding may become less about syntax and more about a shared workflow where human oversight remains.” For all PYMNTS AI coverage, subscribe to the daily AI Newsletter. Recommended Anthropic Debuts Security Tool as Cybersecurity Stocks Fall Gemini Cuts More Staff as Crypto Firm's Valuation Plunges 82% Wellspring Debuts Treasury Management Tool for Small Businesses Google Exec Warns AI Wrapper Startups Could Be in Trouble SEE MORE IN: AI, ANTHROPIC, CLAUDE, CYBERSECURITY, NEWS, PYMNTS NEWS, VIBE CODING, WHAT'S HOT