AI Shifts Cybersecurity Focus from Finding Flaws to Fixing Them
- By John K. Waters
- 06/22/26
One of cybersecurity’s most difficult challenges has always been discovering vulnerabilities before enemies do. A growing number of security specialists now state artificial intelligence is altering that equation, moving the focus from finding defects to repairing them rapidly enough to prevent exploitation.
At the center of that shift is Anthropic’s Task Glasswing, a cybersecurity initiative that provides picked organizations with access to Claude Mythos Preview, an advanced AI model developed to recognize software vulnerabilities and possible attack courses.
According to Anthropic, Job Glasswing has expanded to more than 150 organizations across more than 15 countries and has helped determine more than 10,000 high- or critical-severity vulnerabilities in participating companies and software application jobs. The figures were revealed by the company in materials describing the initiative.
The program gained extra attention just recently when BT Group ended up being the very first U.K. business to publicly sign up with the effort. According to BT and reporting by TechRadar, the telecommunications company plans to utilize the technology to reinforce defenses throughout its networks and consumer systems. BT stated it presently obstructs approximately 4 million cyber attacks each day.
The advancement reflects a wider trend in which AI companies are placing sophisticated designs as cybersecurity tools for governments, critical facilities operators, and big enterprises.
According to Anthropic, Project Glasswing individuals include companies from sectors such as telecommunications, health care, energy, and government. The business likewise notes major technology and financial companies, including Microsoft, Google, Apple, Nvidia, Amazon Web Solutions, CrowdStrike, Cisco, and JPMorgan Chase as individuals or partners in cybersecurity-related efforts.
Security specialists say one of AI’s most substantial advantages is its ability to evaluate big code bases quickly and determine relationships amongst vulnerabilities that might be hard for human experts to acknowledge.
According to reporting by The Wall Street Journal, executives at Visa included with the effort stated the Claude Mythos Preview design can link multiple lower-severity vulnerabilities into reasonable attack chains, helping protectors recognize threats that may otherwise go unnoticed.
The guarantee of the innovation, nevertheless, is accompanied by issues about misuse.
Anthropic has mentioned that Claude Mythos Preview is not broadly offered due to the fact that the exact same abilities that can help defenders identify vulnerabilities might potentially help assailants in locating weaknesses more effectively. The business has stated gain access to is limited to vetted companies since of concerns that advanced cybersecurity designs might be utilized for offensive purposes if released widely.
Those concerns have ended up being significantly popular as federal governments and regulators examine the security ramifications of frontier AI systems. Anthropic has positioned Project Glasswing as a protective cybersecurity initiative while keeping restrictions on public access to the underlying model.
The outcome is a growing debate over whether AI’s greatest influence on cybersecurity will come from strengthening defenses, making attacks more advanced, or both.
For now, supporters of the innovation argue that AI is assisting companies attend to a longstanding issue: the failure to identify and remediate vulnerabilities quick enough. If the technology continues to enhance, cybersecurity specialists say the bottleneck may no longer be finding defects, but identifying which ones to fix initially.
Updates on Task Glasswing are readily available here on the Anthropic site.
About the Author
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He’s been blogging about innovative innovations and culture of Silicon Valley for more than 20 years, and he’s written more than a dozen books. He also co-scripted the documentary Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at [e-mail protected]