The share of biology preprints acknowledging large language model (LLM) use has grown exponentially since 2023. In three years, LLM use acknowledgments have nearly caught up to acknowledgments of specialized AI tool use (17% versus 19%), which have climbed steadily since 2021.
| Month | bioRxiv preprints examined | Used a specialized AI tool | Acknowledge LLM use (any purpose) | Acknowledge LLM use for research | Share using a specialized AI tool (%) | Share acknowledging LLM use (%) | Share acknowledging LLM use for research (%) |
|---|---|---|---|---|---|---|---|
| 2020-01 | 2,657 | 38 | 1 | 0 | 1.43 | 0.04 | 0.00 |
| 2020-02 | 2,744 | 46 | 0 | 0 | 1.68 | 0.00 | 0.00 |
| 2020-03 | 3,036 | 48 | 0 | 0 | 1.58 | 0.00 | 0.00 |
| 2020-04 | 3,413 | 38 | 1 | 0 | 1.11 | 0.03 | 0.00 |
| 2020-05 | 3,797 | 48 | 4 | 4 | 1.26 | 0.11 | 0.11 |
| 2020-06 | 3,819 | 51 | 2 | 2 | 1.34 | 0.05 | 0.05 |
| 2020-07 | 3,696 | 61 | 4 | 3 | 1.65 | 0.11 | 0.08 |
| 2020-08 | 3,186 | 50 | 0 | 0 | 1.57 | 0.00 | 0.00 |
| 2020-09 | 3,151 | 44 | 5 | 5 | 1.40 | 0.16 | 0.16 |
| 2020-10 | 3,188 | 56 | 2 | 2 | 1.76 | 0.06 | 0.06 |
| 2020-11 | 3,006 | 60 | 2 | 2 | 2.00 | 0.07 | 0.07 |
| 2020-12 | 2,994 | 71 | 4 | 4 | 2.37 | 0.13 | 0.13 |
| 2021-01 | 3,002 | 57 | 2 | 2 | 1.90 | 0.07 | 0.07 |
| 2021-02 | 3,156 | 65 | 0 | 0 | 2.06 | 0.00 | 0.00 |
| 2021-03 | 3,523 | 78 | 2 | 2 | 2.21 | 0.06 | 0.06 |
| 2021-04 | 3,247 | 81 | 3 | 2 | 2.49 | 0.09 | 0.06 |
| 2021-05 | 3,229 | 77 | 4 | 3 | 2.38 | 0.12 | 0.09 |
| 2021-06 | 2,993 | 92 | 1 | 1 | 3.07 | 0.03 | 0.03 |
| 2021-07 | 2,927 | 91 | 0 | 0 | 3.11 | 0.00 | 0.00 |
| 2021-08 | 2,847 | 102 | 3 | 2 | 3.58 | 0.11 | 0.07 |
| 2021-09 | 2,956 | 130 | 1 | 1 | 4.40 | 0.03 | 0.03 |
| 2021-10 | 3,058 | 111 | 6 | 4 | 3.63 | 0.20 | 0.13 |
| 2021-11 | 2,870 | 160 | 3 | 2 | 5.57 | 0.10 | 0.07 |
| 2021-12 | 2,965 | 154 | 4 | 3 | 5.19 | 0.13 | 0.10 |
| 2022-01 | 2,855 | 156 | 1 | 0 | 5.46 | 0.04 | 0.00 |
| 2022-02 | 2,646 | 138 | 3 | 2 | 5.22 | 0.11 | 0.08 |
| 2022-03 | 3,189 | 192 | 3 | 2 | 6.02 | 0.09 | 0.06 |
| 2022-04 | 2,585 | 184 | 6 | 4 | 7.12 | 0.23 | 0.15 |
| 2022-05 | 3,087 | 193 | 5 | 4 | 6.25 | 0.16 | 0.13 |
| 2022-06 | 2,890 | 193 | 6 | 4 | 6.68 | 0.21 | 0.14 |
| 2022-07 | 2,950 | 201 | 1 | 1 | 6.81 | 0.03 | 0.03 |
| 2022-08 | 2,905 | 234 | 6 | 3 | 8.06 | 0.21 | 0.10 |
| 2022-09 | 3,283 | 226 | 5 | 4 | 6.88 | 0.15 | 0.12 |
| 2022-10 | 3,211 | 255 | 9 | 6 | 7.94 | 0.28 | 0.19 |
| 2022-11 | 2,993 | 265 | 6 | 3 | 8.85 | 0.20 | 0.10 |
| 2022-12 | 3,010 | 285 | 8 | 5 | 9.47 | 0.27 | 0.17 |
| 2023-01 | 2,987 | 267 | 9 | 4 | 8.94 | 0.30 | 0.13 |
| 2023-02 | 2,995 | 257 | 5 | 1 | 8.58 | 0.17 | 0.03 |
| 2023-03 | 3,447 | 310 | 17 | 8 | 8.99 | 0.49 | 0.23 |
| 2023-04 | 2,953 | 293 | 16 | 7 | 9.92 | 0.54 | 0.24 |
| 2023-05 | 3,104 | 299 | 11 | 5 | 9.63 | 0.35 | 0.16 |
| 2023-06 | 3,266 | 316 | 28 | 10 | 9.68 | 0.86 | 0.31 |
| 2023-07 | 3,177 | 319 | 21 | 12 | 10.04 | 0.66 | 0.38 |
| 2023-08 | 3,297 | 334 | 22 | 3 | 10.13 | 0.67 | 0.09 |
| 2023-09 | 3,586 | 359 | 34 | 11 | 10.01 | 0.95 | 0.31 |
| 2023-10 | 3,348 | 340 | 28 | 9 | 10.16 | 0.84 | 0.27 |
| 2023-11 | 3,623 | 420 | 44 | 18 | 11.59 | 1.21 | 0.50 |
| 2023-12 | 3,364 | 395 | 41 | 13 | 11.74 | 1.22 | 0.39 |
| 2024-01 | 3,222 | 354 | 50 | 10 | 10.99 | 1.55 | 0.31 |
| 2024-02 | 3,372 | 369 | 52 | 14 | 10.94 | 1.54 | 0.42 |
| 2024-03 | 3,746 | 427 | 61 | 14 | 11.40 | 1.63 | 0.37 |
| 2024-04 | 3,286 | 407 | 66 | 22 | 12.39 | 2.01 | 0.67 |
| 2024-05 | 3,576 | 407 | 69 | 23 | 11.38 | 1.93 | 0.64 |
| 2024-06 | 3,670 | 502 | 80 | 24 | 13.68 | 2.18 | 0.65 |
| 2024-07 | 3,712 | 507 | 60 | 18 | 13.66 | 1.62 | 0.48 |
| 2024-08 | 3,468 | 445 | 80 | 19 | 12.83 | 2.31 | 0.55 |
| 2024-09 | 4,002 | 525 | 84 | 15 | 13.12 | 2.10 | 0.37 |
| 2024-10 | 4,049 | 573 | 109 | 29 | 14.15 | 2.69 | 0.72 |
| 2024-11 | 3,685 | 536 | 99 | 22 | 14.55 | 2.69 | 0.60 |
| 2024-12 | 3,837 | 583 | 121 | 31 | 15.19 | 3.15 | 0.81 |
| 2025-01 | 3,703 | 548 | 141 | 32 | 14.80 | 3.81 | 0.86 |
| 2025-02 | 3,646 | 512 | 145 | 44 | 14.04 | 3.98 | 1.21 |
| 2025-03 | 4,102 | 627 | 159 | 39 | 15.29 | 3.88 | 0.95 |
| 2025-04 | 3,837 | 592 | 155 | 44 | 15.43 | 4.04 | 1.15 |
| 2025-05 | 4,331 | 671 | 184 | 49 | 15.49 | 4.25 | 1.13 |
| 2025-06 | 3,946 | 622 | 208 | 58 | 15.76 | 5.27 | 1.47 |
| 2025-07 | 4,049 | 615 | 205 | 55 | 15.19 | 5.06 | 1.36 |
| 2025-08 | 3,915 | 591 | 235 | 59 | 15.10 | 6.00 | 1.51 |
| 2025-09 | 4,769 | 743 | 305 | 57 | 15.58 | 6.40 | 1.20 |
| 2025-10 | 4,736 | 779 | 261 | 50 | 16.45 | 5.51 | 1.06 |
| 2025-11 | 3,884 | 657 | 264 | 63 | 16.92 | 6.80 | 1.62 |
| 2025-12 | 4,388 | 758 | 327 | 59 | 17.27 | 7.45 | 1.34 |
| 2026-01 | 3,924 | 653 | 295 | 59 | 16.64 | 7.52 | 1.50 |
| 2026-02 | 3,993 | 717 | 351 | 80 | 17.96 | 8.79 | 2.00 |
| 2026-03 | 4,702 | 802 | 419 | 94 | 17.06 | 8.91 | 2.00 |
| 2026-04 | 4,265 | 789 | 465 | 105 | 18.50 | 10.90 | 2.46 |
| 2026-05 | 5,164 | 931 | 609 | 117 | 18.03 | 11.79 | 2.27 |
| 2026-06 | 4,496 | 787 | 532 | 121 | 17.50 | 11.83 | 2.69 |
| 2026-07 | 4,674 | 847 | 629 | 130 | 18.12 | 13.46 | 2.78 |
| 2026-08 | 4,268 | 854 | 706 | 143 | 20.01 | 16.54 | 3.35 |
| 2026-09 | 4,922 | 932 | 846 | 190 | 18.94 | 17.19 | 3.86 |
Specialized AI tools include models for protein structure and design, image analysis, single-cell sequencing, and genomics. The most commonly acknowledged is AlphaFold, which predicts protein structure. It became publicly available in July 2021 and is acknowledged in about half of all preprints acknowledging specialized AI tool use. LLM usage is more varied, including literature review, writing, and coding assistance. About 4% of biology preprints in September 2026 acknowledged LLM use for research or analysis, up from about 1% at the start of the year.
Epoch's work is free to use, distribute, and reproduce provided the source and authors are credited under the Creative Commons BY license.
Learn more about this graph
This data insight is based on our AI use in research explorer. We use GPT-5.6 Sol to check every preprint on bioRxiv from January 2020 until September 2026 for acknowledgments of AI use, counting LLM use and specialized AI tool use separately. The OpenAI Decisions API, based on GPT-6 Luna, further classifies LLM use acknowledgments into four use-case categories: research, code and computation, writing, and literature review. Research use includes conceptual contributions such as experimental design, methodological contributions such as data labeling, and analysis, such as statistical analysis. See the full methodology for more information.
Assumptions and limitations
- We measure voluntary AI acknowledgments, which reflect both AI usage and disclosure norms. Our figures may underestimate true AI usage.
- We look at AI use acknowledgments in biology preprints. Preprints may not be representative of peer-reviewed papers.
- AI use acknowledgments and LLM use case classifications are performed by AI models and are therefore subject to error.
- We compiled a list of 150 specialized AI tools by sampling recent preprints on bioRxiv and scanning for mentions of AI tools. This may not be a complete set of specialized AI tools used in biology research.
See the limitations section of the full methodology for more details.
Download this data
Explore this data
Our estimates of the share of papers acknowledging large language model use in physics, biology, statistics, and math.
Related insights
