Now is the time to give LLMs access to the ACM digital library

The dominant read was that ACM is acting like a broker for work it did not create. Several researchers objected that authors already hand over rights under a coercive publishing system, usually without real bargaining power or royalties, and now publishers want to sell AI access on top. That looked hypocritical to many because ACM still makes discovery and human access worse than it needs to be, while suddenly getting ambitious about machine access. A recurring complaint was that "open access" in academic publishing often means shifting costs onto authors or universities, not genuinely widening access. A second thread cut through the legal debate with a practical point. Major labs have probably already scraped most of this material, whether from ACM, author copies, institutional mirrors, or the open web. In that framing, strict licensing mainly punishes actors willing to play by the rules and strengthens incumbents that already grabbed the data. That is why several people landed on a permissive stance for open-weight models in particular, even if they disliked closed AI companies capturing the value. The copyright argument never really settled because it is still unsettled in the courts. Some commenters insisted LLM training and outputs are plainly infringing when they reproduce or substitute for original text. Others argued copyright covers expression, not ideas, and that model use is closer to reading, indexing, or statistical processing than republication. The more grounded takeaway was narrower: the legal fight is real, but the business fight is more immediate. Publishers want to turn training access into a licensable product, while many authors and readers want the benefit to flow to the public, smaller model builders, or the researchers whose work created the corpus in the first place. Underneath all of it was a blunt usability point. People already use Google Scholar and now AI assistants to navigate the literature because publisher platforms are poor search tools. For some, LLMs have already made academic work more reachable by finding primary sources and tracing claims back through citations. That made ACM's pitch sound less like a breakthrough for science and more like an attempt to charge rent on a workflow that has already moved elsewhere.