Computer Science · Information Systems

Information and Cyber Security

Better-trained AI finds more flaws in unfamiliar software projects

AI vulnerability detectors often perform less well on software projects outside their training data. A study finds that cleaner, more varied training data and an encoder-based design improved detection, including a reported 6.8% gain in recall on the BigVul benchmark.

2 days ago·34 views

Abstract editorial illustration accompanying the article: Cybersecurity risk write-ups in filings predict profits, but where they appear matters

AI & Computing

Cybersecurity risk write-ups in filings predict profits, but where they appear matters

A study of 13,800 firm-years examined how companies’ required cybersecurity disclosures vary by topic and where they appear. It found that risk-factor write-ups were more consistently linked to next-year profitability and valuation than similar text in management’s discussion section.

3 weeks ago·19 views