Researchllmpassword securityirregularpasskeys
AI Generates Predictable Passwords Exposing Security Risk
7.2
Irregular, an AI cybersecurity firm, found that ChatGPT, Claude, and Gemini produced repetitive, predictable passwords in tests reported Feb. 26, 2026. Claude reused one password 10 times among 50 outputs and Gemini often returned similar three-noun-plus-digits structures, reducing effective entropy. The article advises using passkeys, two-factor authentication, or manual/randomized methods instead of relying on LLM-generated passwords.
Key Points
- 1Show repetitive outputs: Claude reused 'K9#mPx$vL2nQ8wR' ten times among 50 generated passwords
- 2Indicate LLMs lack true randomness, making 'strong-looking' passwords structurally predictable and narrower in search space
- 3Recommend alternatives: teach users password-creation methods, use passkeys or 2FA instead of AI-generated passwords
Scoring Rationale
Practical security finding with actionable alternatives, boosted by clear tests, but limited by single-firm methodology and modest novelty.
Sources
Public references used for this report.
Practice with real Logistics & Shipping data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
High-Value Overnight OrdersEasyDelivered International ShipmentsMediumOn-Time Delivery Rate by CarrierHard
250 free problems · No credit card
See all Logistics & Shipping problems
