How are Colleges Adapting Admissions Processes and Exam Proctoring to Combat AI Cheating in Application Essays and Standardized Testing?

The proliferation of advanced generative artificial intelligence has fundamentally altered how higher education institutions evaluate prospective and current students. Because AI models can instantly generate highly polished essays and accurately answer complex test questions, traditional take-home assignments and unmonitored remote exams are no longer reliable indicators of a student’s individual capability.

In response, colleges and universities have shifted their focus away from relying solely on AI text-detection software, which has historically struggled with false positives and evasion tactics. Instead, institutions are restructuring their admissions frameworks and testing environments to emphasize process verification, live assessments, and advanced proctoring technologies to ensure the authenticity of high-stakes student work.

Adapting the Admissions Process

To verify that application materials accurately reflect a student’s abilities, admissions departments are introducing new friction points and verification steps into the application lifecycle.

• Proctored Writing Samples: Many institutions are moving away from the traditional take-home personal statement. Instead, applicants are required to complete timed, supervised writing assessments within a secure browser or at a designated testing center.
• Video Interviews: Colleges are increasing their reliance on live or asynchronous video interviews. These sessions allow admissions officers to gauge a student’s spontaneous communication skills and verify that their verbal articulation aligns with the persona presented in their written application.
• Draft Tracking and Version History: Some application portals now require students to draft their essays directly within a proprietary platform. These platforms track the writing process, capturing keystroke dynamics, revision history, and time spent, ensuring the text was generated organically rather than pasted from an external AI tool.
• Hyper-Specific Prompts: Admissions committees are designing essay prompts that require highly localized, deeply personal, or multimedia-linked reflections. These prompts are specifically engineered to be difficult for AI to answer convincingly without extensive, highly specific human prompting.

Evolving Exam Proctoring and Standardized Testing

Standardized testing organizations and university testing centers have implemented stringent technological and environmental controls to secure the integrity of exams.

• Behavioral Biometrics: Modern remote proctoring software now analyzes typing patterns, mouse movements, and navigation habits. By establishing a baseline for how a specific student interacts with their device, the software can flag anomalies that may indicate a student has handed control over to a third party or an automated script.
• Enhanced Environment Scanning: Remote proctoring protocols now frequently require 360-degree camera sweeps of the testing room. Additionally, some platforms mandate a secondary camera angle, such as a smartphone positioned behind the student, to ensure no unauthorized devices or AI assistants are being accessed off-screen.
• Secure Browser Environments: Institutions mandate the use of locked-down testing browsers. These applications take control of the operating system to block access to external websites, disable clipboard copying and pasting, and shut down background processes, preventing AI tools from running concurrently with the exam.
• Return to In-Person Testing: For the highest-stakes exams, there is a notable shift back to physical testing centers. In these environments, variables can be strictly controlled, network access is restricted, and physical proctors can actively monitor for the use of unauthorized smart devices.

The Shift from Detection to Verification

The overarching strategy in higher education has transitioned from attempting to catch AI-generated text after the fact to verifying the human effort during the creation process.

• Authentication over Detection: Because AI models continually evolve to bypass text detectors, colleges prioritize proof of the student’s workflow. Verifying identity and monitoring the environment is now considered more effective than analyzing the final submitted text.
• Holistic Review Adjustments: Admissions rubrics are being adjusted to place less weight on the structural polish of an essay. For example, Duke University stopped assigning numerical ratings to applicant essays as part of its holistic review process, a change made in direct response to the rise of generative AI. Greater weight is now placed on verified metrics, such as proctored exam scores, live interview performance, and documented extracurricular achievements.

Summary

To combat the use of generative AI in admissions and testing, colleges are fundamentally changing the environments in which assessments occur. By implementing secure browsers, behavioral biometrics, live interviews, and proctored writing sessions, higher education institutions are moving away from trust-based, unmonitored submissions. The focus is now firmly on verifying the authenticity of the student’s process and ensuring that high-stakes evaluations accurately reflect individual human capability.