Developments in AI create increasing challenges for security AND humans
Research
published in Smashing Magazine confirms what many developers have suspected: CAPTCHAs are now blocking more humans than bots. As AI-driven tools have become faster and more accurate at completing these tests, the barriers designed to keep bad actors out have become increasingly frustrating for real users.
The problem is sharper for people with disabilities. Screen readers used by visually impaired people cannot interpret visual CAPTCHAs, and the distorted text versions are effectively unusable. Newer formats involving maths problems or puzzles create additional barriers for people with dyslexia, dyscalculia or cognitive impairments.
When clients ask us to prevent bot spam, fraudulent sign-ups or unwanted data scraping, we now recommend Google’s reCAPTCHA v3 or Cloudflare Turnstile. Although not completely bulletproof, both use invisible verification methods that stop most bots without interrupting genuine users, stepping in with additional checks only when something looks suspicious.
