bapt@ created a survey for FreeBSD developers on the use of AI and LLMs. He presented the results to core@, and a summary was shared with developers. 140 developers responded.

• Policy: 88% want the Project to adopt a policy, either a formal, written policy (49%) or light guidance and best practices (39%). 6% said that AI should be explicitly discouraged, and 3% that the existing contribution rules are sufficient.

• AI-assisted code: 35% said it must be disclosed, 25% that it must be reviewed more strictly than human code, 22% that it should be treated like any other contribution, and 16% that it should be rejected outright.

• Safeguards: mandatory human review (78%) and mandatory disclosure (69%) were requested most often, followed by provenance or licensing documentation (47%) and mandatory testing (39%).

• Disclosure: 84% want AI use disclosed, either always (48%) or when its involvement is significant (36%). 79% want AI-assisted commits marked in the commit log, and 56% prefer a tag that names the tool.

• Concerns: the main concerns about quality are correctness and subtle bugs (39%), licensing and provenance (26%), and maintainability (21%). 75% are somewhat or very concerned about copyright and licensing.

• Principles: accountability, meaning that a human remains responsible (91%), and transparency (78%) were selected most often, and preserving code quality was the top-ranked goal for a policy.

• Other uses: 65% would allow AI-generated documentation with human review. 54% would allow AI tools on Project infrastructure, most of them with restrictions, and 29% would not.

https://www.freebsd.org/status/report-2026-07-2026-09/#_freebsd_core_team