Ars Technica published its formal AI policy: reporters may use vetted AI tools for research but cannot attribute AI-generated or AI-summarized material to named sources.
Key Takeaways
AI tools are permitted for navigating large document volumes, summarizing background material, and searching datasets, but all sourced claims require direct reporter review.
No AI-generated quotes, paraphrases, or characterizations of named sources are allowed, overriding any use of AI in the research phase.
AI-generated images, audio, and video cannot be published as authentic documentation; synthetic media in AI coverage must be clearly labeled near the material.
Every reporter using AI in a story must disclose it to editors and retains full personal accountability – tools and colleagues cannot absorb responsibility.
The policy was last updated April 22, 2026, and Ars states violations have already led to action.
Hacker News Comment Review
The policy is widely read as a direct response to a prior incident where a reporter was fired over fabricated AI-generated quotes; commenters note several rules are narrowly scoped to that specific failure mode rather than systemic AI risk.
Skeptics flag a structural contradiction: permitting AI for research summarization while banning AI-summarized attribution creates an ambiguous middle layer that editors must police case by case.
A broader concern raised is that AI content generation degrades the original-content ecosystem that LLMs depend on for training data, threatening the long-term quality of both journalism and AI outputs.
Notable Comments
@legitster: flags the bootstrap problem – AI needs original human content to train on, but mass AI generation erodes the incentive to produce it.
@applfanboysbgon: argues the permitted research-summarization use is itself unreliable given LLM hallucination behavior, making the policy self-contradicting at its foundation.
