Security Evaluations of GitHub's Copilot
Bibliographic record
Abstract
Code generation tools driven by artificial intelligence have recently become more popular due to advancements in deep learning and natural language processing that have increased their capabilities. The proliferation of these tools may be a double-edged sword because while they can increase developer productivity by making it easier to write code, research has shown that they can also generate insecure code. In this thesis, we perform two evaluations of one such code generation tool, GitHub's Copilot, with the aim of obtaining a better understanding of their strengths and weaknesses with respect to code security. \nIn our first evaluation, we use a dataset of vulnerabilities found in real world projects to compare how Copilot's security performance compares to that of human developers. In the set of (150) samples we consider, we find that Copilot is not as bad as human developers but still has varied performance across certain types of vulnerabilities. In our second evaluation, we conduct a user study that tasks participants with providing solutions to programming problems that have potentially vulnerable solutions with and without Copilot assistance. The main goal of the user study is to determine how the use of Copilot affects participants' security performance. In our set of participants (n=21), we find that access to Copilot accompanies a more secure solution when tackling harder problems. For the easier problem, we observe no effect of Copilot access on the security of solutions. We also capitalize on the solutions obtained from the user study by performing a preliminary evaluation of the vulnerability detection capabilities of GPT-4. We observe mixed results of high accuracies and high false positive rates, but maintain that language models like GPT-4 remain promising avenues for accessible, static code analysis for vulnerability detection. \nWe discuss Copilot's security performance in both evaluations with respect to different types of vulnerabilities as well its implications for the research, development, testing, and usage of code generation tools.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.006 | 0.032 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".