Synthesis history

The main substantive addition is Ars Technica's May 19 overview [^7706] contextualizing Co-Scientist alongside FutureHouse's simultaneously published Nature paper — introducing FutureHouse as a competing system with dis…

A Columbia Nursing AI-assisted audit published in The Lancet — auditing 2.5 million biomedical papers and finding nearly 3,000 with fabricated citations [^20252][^20250] — escalates the research integrity concern from s…

CIDRAP (Center for Infectious Disease Research and Policy) has published an independent review of rising fake reference rates in biomedical papers [^19713], corroborating Retraction Watch's figures from outside the misc…

Two Retraction Watch reports quantify the integrity risks that were previously abstract: 1 in 277 PubMed papers in 2026 shows fabricated references [^18878] and illicit AI use appears in hundreds of peer reviews [^18879…

Three substantive additions this pass: (1) Nature Communications published a peer-reviewed paper 'Risks of AI scientists: prioritizing safeguarding over autonomy' [^17316], moving the critique of AI autonomy from editor…

Most new items this pass are amplification without new substantive claims: social media posts about the 'Towards end-to-end automation of AI research' Nature paper [^15841][^15842], Google I/O roundup coverage [^15403][…

Three significant developments emerge this pass. First, Nature itself published a commentary titled 'Why AI cannot do good science without humans' [^15479] alongside the research papers — the first peer-level institutio…

The main new developments this pass are the broadening of media coverage to mainstream tech outlets (The Verge, Engadget, CNET, Mashable) via Google I/O 2026 [^11684][^11685][^11686][^11687], ERA's EurekAlert science-pr…

The main new developments this pass are the identification of ERA (Empirical Research Assistance) as the third concurrent Nature paper, with a dedicated Google Research blog post framing it as an AI system for writing e…

The Co-Scientist Nature paper was formally published on May 19, 2026 [^8267][^8269], moving the system's claims into peer-reviewed record and addressing a prior open question. A concurrent tweet thread notes three Natur…

Google DeepMind launched Co-Scientist, a multi-agent AI research partner, alongside the broader Gemini for Science platform in mid-May 2026 [^7301]. The rollout was accompanied by five validated case studies spanning li…