Evidence · DiggingBeagle record
The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
The VELA study tests 44 known web application flaws and reports that forty were realizable in a scripted baseline; coding agents could exercise all forty at least once in the evaluated conditions.
- Source role
- primary disclosure
Evidence record
The VELA study tests 44 known web application flaws and reports that forty were realizable in a scripted baseline; coding agents could exercise all forty at least once in the evaluated conditions.
Claim-level citations (2)
- supportsVELA shows agent-mediated web access does not eliminate backend vulnerabilities: The VELA study tests 44 known web application flaws and reports that forty were realizable in a scripted baseline; coding agents could exercise all forty at least once in the evaluated conditions.
Abstract, experiment design and interface comparison
- supportsVELA shows agent-mediated web access does not eliminate backend vulnerabilities: The VELA authors report that 91 percent of 44 selected backend vulnerabilities remained exploitable through a reference AI-mediated interface in their controlled benchmark.
Abstract, reference interface exploitability and legitimate workflow tradeoff