Quick take: This morning’s strongest stories all concern power over dependency: who controls a body, what a medicine’s approval really proves, and how much of a digital tool survives when its upstream supplier walks away.
Greenland’s contraception inquiry confirms rights violations without settling the genocide question
Two expert reports released on 28 August agree on the human core of Greenland’s “spiral case”: women and girls were subjected to contraception without valid consent, violating rights to private and family life and, in some cases, protection from degrading treatment. A 2025 joint inquiry documented at least 4,000 IUD insertions or contraceptive injections between 1966 and 1991, including in girls as young as 12. The new reports differ over how systematic the violations were, and neither establishes the specific intent required for a legal finding of genocide. Associated Press and Reuters describe that unresolved divide. The danger is that argument over the strongest legal label eclipses the proven harm or turns survivors into instruments of current geopolitics. For individuals, this is also a warning about consent hidden inside institutional authority. Support survivor-led reconciliation and compensation, and ask whether any future inquiry gives affected women control over testimony, records and remedies—not merely a place in a historical narrative.
Mounjaro gains a heart-risk indication—but the comparison needs careful reading
On 28 August, the US Food and Drug Administration expanded Mounjaro’s label to include reducing cardiovascular death, non-fatal heart attack or non-fatal stroke in adults with type 2 diabetes at high cardiovascular risk. The manufacturer’s announcement and Reuters report describe a trial of more than 13,000 people comparing tirzepatide with Lilly’s older drug Trulicity. Mounjaro’s major-event rate was 8% lower, but the confidence interval crossed the line for statistical superiority; the study’s prespecified success was showing that Mounjaro was not unacceptably worse. The approval is promising because cardiovascular risk is central to diabetes care and treatment choices can now consider glucose, weight and heart outcomes together. It does not show that every patient should switch, and access, gastrointestinal effects and the drug’s boxed thyroid-tumour warning still matter. Ask a clinician for absolute event rates and how the comparator fits your situation, rather than treating “8% lower” as an individual guarantee.
Cursor’s planned OpenAI cutoff reveals the fragility of rented intelligence
OpenAI said on 28 August that it intends to stop supplying models to Cursor after the coding-tool company’s acquisition by SpaceX, with a proposed cutoff on 12 November. OpenAI says it lacks confidence that SpaceX will follow its terms, citing prior experience with Elon Musk’s companies; SpaceX disputes the broader accusations, and Cursor says discussions continue. Those are company positions, not independently established contract violations. The concrete fact is the planned withdrawal, documented in OpenAI’s notice and contextualised by Reuters on 29 August. More competition between model suppliers may make coding tools resilient, and Anthropic says it will expand Claude support. Yet developers can still lose a familiar model because of a dispute above them, while prompts, evaluations and workflow habits may not transfer cleanly. If a team relies on an AI coding assistant, inventory model-specific dependencies now: export reusable instructions, test the same tasks on a second provider, and measure output quality before any forced migration.
