Last Week in AI maps a crowded week around OpenAI and Gemini

Last Week in AI #341 is a roundup, not a primary announcement. Its value is as a signal map: legal pressure around OpenAI, Gemini product competition and research results that deserve separate verification.

Musk lost the lawsuit, Google shipped Gemini updates, OpenAI tackled Erdős

The episode summarizes several stories. The most prominent one is Elon Musk losing his lawsuit against OpenAI, Sam Altman and Greg Brockman. The text says a federal jury in Oakland rejected the 150 billion dollar claim and the judge closed the case because of statutes of limitations.

The second major thread is Google IO 2026. The roundup mentions Gemini app updates and Google trying to compete with ChatGPT and Claude. The title also mentions OpenAI solving Erdős, but that item needs the primary source because the roundup combines multiple links and the detail is not a full article on its own.

The AI market is moving on three fronts at once

The value of roundups is aggregation, not a final verdict. This one shows the AI market moving on three fronts at once: legal scrutiny of old commitments, product competition for the user interface and research results that shape lab reputation.

For product people and founders, the important move is separating noise from risk. The legal case can affect the narrative and governance around OpenAI. Gemini product updates decide whether Google can turn models into a daily work tool.

Every roundup item needs primary verification

A roundup is not the source of truth. It combines secondary sources and short summaries, so every major item needs primary verification. That is especially true for court outcomes and financial claims, where original reporting or legal documents matter.

It is also wrong to force one grand thesis just because everything happened in the same week. Musk's lawsuit, the Gemini app and an OpenAI math result share the AI market context, but they have different causes and consequences.

The practical Gemini test: daily tool, not conference demo

For OpenAI, watch whether the legal win actually clears the way for structural changes, financing or IPO preparation. For Gemini, the practical test is simple: whether people use it as a daily work channel, not only as a conference demo.

For research items inside roundups, keep only the ones with a paper, reproducible result or clear application. Everything else belongs in Radar as a signal, not as a finished conclusion.

Lilith's verdict

A crowded pinboard where a judge, Google product team and OpenAI researchers each pin their own note. There is no single grand thesis about the AI market behind it.