Daily AI digest, October 30

Felix Laurent 30 Oct 2026 4 min read

News

The biggest AI news of October 30 was about plans and limits. OpenAI dropped its planned ChatGPT Pro cut and doubled usage limits instead. Anthropic renamed its top Claude plan to Ultra and removed the 5-hour limit for verified accounts. Google ran Gemma 4 on a TPU in orbit for the first time. Below are the big stories of Friday, October 30, plus the late Thursday news that landed after our last digest.

OpenAI drops the Pro cut and doubles limits

OpenAI had planned to cut usage for existing ChatGPT Pro 200 subscribers from October 30 while keeping the $200 price. We reported that plan in yesterday's digest and in a separate article.

Late on Thursday, OpenAI reversed it. Tibo, who speaks for ChatGPT and Codex on X, wrote that OpenAI "won't be reducing usage limits for Pro 200". He also announced that OpenAI is doubling all usage limits for the Pro 100, 200 and 500 tiers and Business Premium, effective immediately. The change mostly covers ChatGPT Work and Codex, which share one allowance, BleepingComputer reports. OpenAI gave no exact numbers for any tier.

Anthropic renames Max 20x to Ultra and drops the 5-hour limit

Anthropic renamed its $100 Max 5x plan to Max and its $200 Max 20x plan to Ultra, Claude said on X. Prices stay the same. The bigger change is in the limits: Ultra has no 5-hour limit for verified accounts, starting with most existing subscribers.

The new pricing page lists Ultra at 10x Pro usage per week. It also includes $200 a month in API credits, up from $100 on the old Max plan. Max stays at 5x Pro usage every 5 hours and every week. Weekly limits still apply on both plans.

Both moves were about heavy users. OpenAI and Anthropic now compete on how much work their top plans allow, not only on the price.

Google runs Gemma 4 on a TPU in orbit

Google says its Project Suncatcher prototype satellite returned the first AI response from a data-center-class TPU in space. The first prompt asked Gemma 4, running on board, to write a haiku about being in orbit. Google calls it a "Hello World" milestone for space-based AI compute.

The satellite launched on October 1 on a SpaceX Falcon 9 rideshare, built with Planet. It carries four TPUs but can run only one at a time, for about 20 minutes, on up to 1 kW of power, TechCrunch reports. Google plans two more satellites with newer chips in 2027. Our breakdown of where the reports differ is here.

Regulators and courts

Two moves on October 30 were about who may use publishers' content for AI.

The UK Competition and Markets Authority opened an investigation into Meta's new commercial terms for publishers and businesses. The terms would let Meta use publisher content for its AI features. The CMA says they could amount to unfair trading conditions for thousands of UK businesses, and the changes are now on hold. Meta says there are no grounds for an investigation.
The Wikimedia Foundation filed an amicus brief in the combined copyright case against OpenAI, which covers 12 lawsuits including The New York Times'. It argues that training models can be fair use, but that retrieval-augmented generation which returns Wikipedia content should credit it under CC BY-SA 4.0.

Both cases test whether AI products can reuse content without paying or crediting the source.

Jobs and money

Four stories showed how AI is changing jobs and the chip supply chain.

Comcast is laying off 634 employees at its Philadelphia headquarters and 360 more in New Jersey, Washington state and California. It is deploying AI agents for customer contact. The cuts take effect December 31, and Comcast plans more than 1,100 new hires, The Philadelphia Inquirer reports.
Goldman Sachs found that workers displaced by AI take about a month longer to find a new job. They are about 10 percentage points more likely to still be unemployed a year later, Fortune reports.
AUO will sell its Taichung Phase 2 display plant to TSMC for TWD 38.8B, about $1.2B. The site covers about 429,000 square meters, TrendForce reports; reports say TSMC plans to use it for CoWoS advanced packaging of AI chips.
Crosby, an AI-first law firm that has raised about $85M, bid for Weil Gotshal's trademark, brand and some internal data. The 1,200-lawyer firm is preparing to vote on dissolution, TechCrunch reports.

Comcast's cuts and Goldman's data describe the same shift from two sides.

What the day was about

Friday was about how much AI people get for their money. OpenAI and Anthropic both loosened limits on their $200 plans within a day of each other. OpenAI doubled its Pro limits, and Anthropic dropped the 5-hour limit on Ultra.

The costs moved elsewhere. Comcast is cutting 994 jobs as it deploys AI agents, while hiring 1,100 people for other roles. Goldman's data shows displaced workers take longer to recover. On October 30 AI got cheaper for its heaviest users, while the price of that shift landed on the people it replaces.