New Intelligence Report
Editors: Aeneas, Dinghui
[Editor's Note] Just now, Anthropic released the code security tool Claude Code Security, which immediately wiped out over $10 billion in market value from security stocks! Cybersecurity leader CrowdStrike's stock plummeted over 6.5% on the spot, throwing the market into extreme panic: Are traditional security tools now completely obsolete?
During the Lunar New Year period, the continuous earthquakes in the AI community haven't stopped.
Yesterday's Gemini 3.1 Pro had just made its debut, and then came another new bombshell—
Just now, Anthropic released a code security tool called Claude Code Security, which can efficiently scan code repositories for vulnerabilities and automatically generate targeted patches, far surpassing traditional tools.
Does this sound like just an ordinary technical update?
As a result, the moment the news broke, U.S. cybersecurity stocks immediately plunged collectively.
CrowdStrike, Cloudflare, Okta and other security stocks instantly fell over 5%, with total market value evaporating by more than $10 billion—overnight, tens of billions in market cap vanished!
Looking closely, cybersecurity leader CrowdStrike once plunged 6.5%, Cloudflare fell over 6%, SailPoint dropped 6.8%, Okta declined 5.7%, Zscaler fell 3.5%, and the Global X Cybersecurity ETF dropped 3.8%, with year-to-date cumulative losses expanding to 14%.
A single AI feature update directly caused a trillion-dollar industry sector to bleed on the spot.
This isn't volatility—this is panic!
Foreign media quickly published articles: Cybersecurity company stocks plunged on Friday, simply because Anthropic introduced a security feature in the new Claude model.
Security Stocks Turn Red Across the Board, Giant Market Caps Shrink by Billions
As soon as Anthropic's new weapon was released, the cybersecurity market tasted blood.
Cybersecurity leader CrowdStrike's stock fell over 6.5%, with Cloudflare and Okta following suit. Total market value evaporated by more than $10 billion overnight.
Global investors fell into panic: AI will directly eat into professional security vendors' market share!
The Cybersecurity ETF fell 4.9%, reaching its lowest point since November 2023 at closing.
In these first few weeks of 2026, market anxiety has been continuously heating up.
The iShares Expanded Tech-Software ETF has fallen more than 23% this year, poised to record its largest quarterly decline since the 2008 financial crisis!
Ironically, Cloudflare was once seen as a beneficiary of Anthropic's rise; in late January this year, its stock even rose significantly after the adoption rate of an open-source AI assistant based on the Claude model increased.
However, markets change in an instant.
Consider this: half of the startups in YC right now are in the security space. With Anthropic releasing this new tool, do they still have a path forward?
Every time Claude releases a new version, this is what goes through the minds of SaaS business owners:
AI is overturning us with overwhelming force!
Claude Becomes the Ultimate White Hat Hacker
Why did Claude Code Security cause the collapse of global security stocks?
The reason is that it's not just a tool, but a complete disruption of traditional security defense systems.
While security teams were still worrying about piles of vulnerability tickets, Claude had already transformed into the "ultimate white hat hacker," armed with the newly released Claude Opus 4.6, diving headfirst into open-source code repositories.
The results were jaw-dropping: It uncovered over 500 epic-level bugs that had been lurking for over a decade!
Remember, these vulnerabilities had been reviewed line-by-line by countless top human experts, yet remained "slipping through the cracks" in the code, living openly.
Dimensional Strike: Traditional Scanning Tools Crash and Burn?
For a long time, the pain points in the security industry have been stark: vulnerabilities are never-ending, and security experts are always in short supply.
Everyone has been using traditional static analysis tools (SAST), but these tools are often based on rigid rule matching.
They're okay for finding hardcoded passwords, but when encountering complex business logic flaws or privilege escalation vulnerabilities, they're completely blind.
The emergence of Claude Code Security has completely changed the rules of the game.
It doesn't do rigid pattern matching; instead, like a human security veteran with ten years of experience, it truly "understands" and "reasons through" your code!
It can deeply understand how various components interact with each other, following the trail to track data flow paths through applications.
Complex vulnerabilities that rule-based scanning tools can't see have nowhere to hide in the face of its godlike logical reasoning.
No Random Changes! Self-Verification Prevents Hallucinations
Everyone worries about AI hallucinations when AI writes code, but what about AI finding vulnerabilities and fixing bugs?
Anthropic is remarkably solid this time. To filter out the headache-inducing "false positives" (false alarms), every uncovered vulnerability must undergo an extremely rigorous "internal multi-stage verification."
Claude forces itself to play both red and blue teams, desperately trying to prove or disprove its own findings.
Those true high-risk vulnerabilities that emerge in the end are pushed to the security dashboard. Not only are automatically generated patches with ratings provided, but also a detailed "confidence index."
Most importantly, Claude maintains the restraint of "only suggesting, not doing." It identifies problems and provides solutions, but the one pressing the merge button is always the human developer.
Impressive Track Record, Red Team Tested
The power of this system didn't happen overnight.
Over the past year-plus, Anthropic's Frontier Red Team has essentially put Claude through grueling high-intensity training.
Not only was it thrown into CTF security competitions to compete with human hackers, but they also partnered with Pacific Northwest National Laboratory to use AI to defend national critical infrastructure.
True knowledge comes from practice. It was this "hell-level" special training that gave Claude Opus 4.6, released earlier this month, such terrifying defensive capabilities.
Those 500+ rooted-out hidden vulnerabilities are the best medals of honor.
Even Anthropic admits: "We use Claude daily to review our own code, with surprisingly good results!"
AI Offense-Defense Battle: Nuclear-Level Arms Race Begins
Without a doubt, this is a historic turning point.
In the near future, the vast majority of the world's code will be scanned by AI repeatedly.
Hackers will certainly frantically use AI to dig for exploitable weaknesses on a massive scale. But as the saying goes, "as virtue rises one foot, vice rises ten"—faster-moving defense teams can use the same or even stronger AI magic to nip vulnerabilities in the bud.
Currently, Claude Code Security has opened a limited research preview for Enterprise and Team customers!
Open-source project maintainers enjoy an expedited "green light" pathway.
Whoever can master this AI security card first will be able to stand invincible in this life-or-death race of computing power and intelligence.
AI Devours the Middle Layer: What Capital Markets Fear Most Has Arrived!
Over the past two years, AI writing code has already made programmers nervous. But people were still comforting themselves: AI can only write code and fix bugs, but it couldn't replace humans in highly specialized fields like security.
Now, Claude has started scanning for vulnerabilities directly.
This means AI is no longer just generating content, but entering the core workflows of enterprise security, directly targeting the sky-high profit pools of enterprise-grade security services!
What capital markets fear most isn't technology—it's pricing power being shaken.
Cybersecurity companies have long commanded high valuations precisely because offense-defense confrontation is complex and security experts are scarce, making their services highly specialized.
Now with Claude Code Security, a single AI model can complete 80% of vulnerability scanning and repair recommendations. Enterprises only need a few security engineers.
So, do enterprises still need to pay high subscription fees to security vendors?
This is why, even though Claude can't replace CrowdStrike yet, it has already triggered a market selloff.
When investors start asking "Will we still need so many security companies in five years?", the stock market begins to collapse!
The scenario of Anthropic triggering a $1 trillion selloff with Claude seems like it was just yesterday.
Looking back at the series of nuclear-level events triggered by Claude Code since late last year, we can see: the cruelest part of this AI revolution is that it prioritizes devouring the middle layer.
Once model accuracy is high enough, SaaS premiums will be compressed, service fees will be reassessed, and valuation logic will be pierced!
Even more frightening is that Anthropic stated this feature is still just a "limited research preview"—not yet fully open, without commercial-scale verification.
Yet stock prices have already plummeted all the way. This shows that the current speed of AI evolution has far outpaced the product iteration speed of traditional software companies.
This is a dangerous signal flare—even more terrifying things may be yet to come.
References: