AI-rewritten: This is a summary of an article from Singularity Hub, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

---

Mathematicians are reacting with mixed emotions to a new release from OpenAI. Dr. Bridson described the atmosphere as conflicting, noting that while some find the solutions exciting, others feel frightened or devastated. He emphasized that the challenge of extracting human understanding from machine formalities will drive greater ambition in this new era. However, Tim Fernholz reports that OpenAI has not yet met the standards set by an advisory group of elite mathematicians. Specifically, there are gaps between how models express solutions in natural language and their formally expressed versions regarding a million-dollar problem.

In biotechnology, the NIH is focusing on cancer vaccines as its next major project. These treatments are unique to each patient because they prompt the immune system to attack specific tumors. The injections teach T cells to target aberrant proteins found only on a patient’s tumor. Meanwhile, in the tech sector, SpaceX has signed a deal to buy cellular spectrum licenses from Grain Management. This move aims to expand Starlink beyond broadband into the wireless market to compete with traditional telecom companies like AT&T and Verizon.

Experts are raising serious concerns about AI safety and future risks. Professor Stuart Russell warns that pursuing large language models could be a $10 trillion mistake if it leads to systems too capable for current safety guarantees. He suggests further investment is throwing good money after bad. Additionally, the UN Secretary General warned on Thursday that neurotechnologies advancing rapidly could allow governments to access or manipulate brains, threatening human rights. AI companies are also preparing for potential large-scale disruptions within six to 12 months, anticipating they may need to shoulder blame if systems go rogue.