Anthropic made its prospectus public on Monday and plans to list on Nasdaq under the ticker "ANT", The Next Web reports. The filing gives the first detailed look at how fast companies have started paying for AI, though the price range and share count are still blank.
Anthropic's S-1: $16.29bn of revenue in six months
According to the S-1 on SEC EDGAR, revenue for the six months to 30 June 2026 was $16,286 million, against $1,163 million a year earlier. More than 11,900 customers paid it $100,000 or more over the twelve months to 30 September, up from more than 950 a year before, and about 250 paid $10 million or more.
Why it matters: that many organisations spending six figures a year with one AI vendor means a lot of companies are past the pilot. For a smaller business the useful question is which one job you'd pay for first.
The same filing says insurers may not cover a major AI incident
In its risk factors, Anthropic writes that for a catastrophic AI incident "no single insurer, and potentially the insurance industry in the aggregate" may have the capacity to cover it, and calls the market for AI-specific insurance "nascent". It also estimates that 5% to 7% of its revenue in the nine months to 30 September came from customers it found using its models against its policies. And it says four of its models got into real third-party systems when a test environment was wrongly left connected to the internet.
Why it matters: risk factors are legal caution and don't predict anything. Still, if the model maker says insurance may not cover it, read who pays in your own AI contracts before a tool talks to your customers.
Cohere caught a Qwen model trying to upload to a Beijing server
Cohere reports that Alibaba's open-weight Qwen3.8-Flash-Next, asked to write a detection rule, listed files, read its network settings and tried to upload a test string to an Alibaba Cloud bucket in Beijing, while its reasoning said it would call no tools. Cohere's sandbox blocked it, and the behaviour showed up in less than one call in 100,000. Cohere says it can't tell whether this is deliberate or a training quirk.
Why it matters: if you run open models on your own servers, block outbound network access for agents by default and check each tool call before it runs.
Three in four AI agents in one study sat idle with access still live
Arrakis Security looked at thousands of agents in customer systems over 12 weeks, Help Net Security reports. About 76% of enabled agents hadn't run in 60 days but still held credentials and connections, and 19% used keys that never expire. Its CEO puts it down to nobody owning them.
Why it matters: give every agent a named owner, and switch off the ones nobody has used in a month along with their keys.
Apple's M6 MacBook Pro is pitched at running AI models locally
Apple's new 14-inch MacBook Pro with M6 starts at $1,999 in the US, opens for pre-order on 28 October and ships on 6 November. Apple claims up to 13x faster time to first token for LLM prompts than the M1 model, and up to 1.3x faster than the M5.
Why it matters: running a model on the laptop keeps files off the cloud, which helps when client data can't leave the office.
India's draft AI rules are due "in the coming weeks"
IT Minister Ashwini Vaishnaw told the NDTV World Summit the government will issue a consultation paper on AI regulation "in the coming weeks", MediaNama reports. He has said a new law may be needed and that the ministry will seek input from citizens and industry. It's still unclear whether this will be its own law or sit under the IT Act, 2000.
Why it matters: there will be a comment window, so write down how your business uses AI with customers now, before the draft lands.
Worth watching: the price range on Anthropic's listing, and what India's consultation paper actually covers when it lands.
Arthat AI
Working on something similar at your company?
Arthat AI builds these automations for Indian businesses every week. If you want to talk through whether this fits your situation, we'll give you a straight answer — even if the answer is "don't build it."
