A sharp market sell-off has exposed growing unease around the colossal sums being poured into AI infrastructure, with chipmakers and tech giants taking heavy hits. Elsewhere, the major AI labs have backed a shared standard to let website owners opt out of web scraping, and an official Anthropic extension was caught harbouring a backdoor.
In this round-up
Wall Street took fright on 30 October 2026 as mounting capital expenditure on artificial intelligence triggered a broad market retreat. Over the past fortnight, Alphabet, Amazon, Meta, and Microsoft outlined plans to spend more than $740 billion this year, largely on chips and data centres. However, cash flow is drying up. Alphabet wiped out $500 billion (around £374 billion) in market value after revealing negative quarterly free cash flow, as its own AI summaries eat into search ad growth. Intel shares dropped 28% after warning of zero positive cash flow and poor PC demand.
Investors are increasingly worried that companies are burning billions on computing power without a clear route to profit. For event businesses, there is no immediate crisis, but the wider context matters. The era of tech firms treating experimental AI products as free loss leaders is meeting cold financial reality. Expect software subscriptions to get stricter and price rises to creep in as enterprise software vendors try to justify their enormous computing bills.
Read more: techcrunch.com · bbc.co.uk · techcrunch.com
OpenAI, Google, Microsoft, and Anthropic have agreed to support a single standard that lets website owners opt out of having their content scraped for AI training. Proposed by an Anthropic engineer through the Internet Engineering Task Force, the mechanism uses a shared policy token inside a website's standard robots.txt file, the basic text file used to instruct search engines.
Previously, web managers had to block each AI company individually, playing an endless game of whack-a-mole as new web crawlers appeared. Under the new approach, one command covers all participating firms. Anthropic has already adopted it, OpenAI plans to follow in the coming months, and Google and Microsoft intend to roll it out next year. Meta expressed support but stopped short of promising compliance.
For event organisers, venues, and agencies, this is a clean, practical win. If your website publishes proprietary conference presentations, white papers, or research you would rather protect from AI models, you will soon be able to ask your web agency for a simple, single-line update without hurting your normal Google search visibility.
Read more: theverge.com
An official extension for Anthropic's Claude coding model was compromised on Cursor, an AI-powered developer tool, exposing up to 5,000 developer machines to a malicious backdoor. The rogue version was distributed directly through Anthropic's official verified account on Cursor for nearly 12 hours on 30 October 2026.
Security researchers found the tampered code attempted to open a persistent reverse connection to an attacker's server, while harvesting environment variables, system information, and public IP addresses. Crucially, stolen variable names allow attackers to identify cloud access keys and API tokens, which are the digital credentials used to connect corporate software. Cursor removed the extension and rotated keys, but Anthropic has yet to issue a direct customer warning.
If your agency or technical production team builds bespoke event apps, attendee portals, or integrations using Cursor and Claude, alert your developers immediately. They must audit developer machines, verify extension versions, and rotate any API credentials or cloud access keys that were stored locally.
Read more: arstechnica.com
Ollie, an AI assistant aimed at family scheduling and personal organisation, has announced it will shut down on 6 November 2026. Chief executive Bill Lennon cited intense competition from rival services and the rising cost of capital, admitting the small team could not keep up with giants in the space.
The closure is notable because Ollie had raised a $7.5 million seed round and made data security its central selling point, becoming one of the first consumer AI apps to secure SOC 2 compliance, an independent standard that audits how carefully a company handles customer data. The company is offering data export instructions and refunds to subscribers.
It is a reminder for event professionals when choosing independent productivity apps. Many smaller niche AI tools that look perfect for managing run sheets or client communications rely on expensive third-party computing, and privacy alone does not guarantee a viable business. Keep sensitive event schedules and client details anchored in enterprise-backed platforms with clear service guarantees rather than niche consumer startups.
Read more: techcrunch.com
In an appearance on the Cheeky Pint podcast, OpenAI chief executive Sam Altman dismissed the idea of running AI models locally on personal computers, arguing that consumer PCs hog scarce memory chips. A global shortage of dynamic random-access memory (DRAM) has left memory manufacturers Samsung and SK Hynix tied into long-term cloud contracts with AI giants, driving up the retail cost of standard consumer hardware.
Altman argued that leaving powerful RAM on individual desks, where it sits idle for 95% of the time, is an inefficient allocation of limited supply. Instead, he believes memory belongs almost entirely in centralised data centres to ensure maximum utilisation.
For event and production teams, the immediate takeaway is practical budgeting. If you are planning an annual equipment refresh for on-site production laptops, graphics workstations, or edit suites, expect higher hardware quotes and tighter inventory. The AI cloud buildout is directly driving up the price of ordinary high-spec computers.
Read more: theverge.com
Apple will release Apple Intelligence in China starting with an iOS 27.3 beta on 4 November 2026, ahead of a wider public launch in December. To satisfy local regulatory approvals secured in July, Apple trained a bespoke large language model specifically for the Chinese market and partnered with Alibaba to integrate its Qwen chatbot, mirroring its ChatGPT arrangement elsewhere.
The release will not include the revamped Siri assistant, focusing instead on text summaries, proofreading, object removal in photos, and real-time translation tools.
For UK agency and exhibition teams producing events in China or welcoming Chinese corporate delegations, this confirms that Western iOS setups cannot rely on standard OpenAI-linked workflows in the region. Dual-market event apps will need to accommodate local Chinese AI integrations like Qwen when catering to mainland attendees.
Read more: theverge.com
Blockchain intelligence firms TRM Labs and Chainalysis have linked North Korea's state-backed Lazarus Group to a $6.3 billion (£4.7 billion) cyber attack on cryptocurrency exchange Binance. Experts noted that attackers automated unauthorized withdrawals from an internet-connected hot wallet and deployed unprecedented laundering patterns, with specialists suggesting artificial intelligence was plausibly used to help infiltrate systems and automate fund transfers.
While high-level crypto heists are far removed from managing venues, the sheer scale of modern automated intrusion serves as a timely prompt for event registration systems. Delegate payment gateways and stored credit card databases remain prime targets for automated exploit bots. Standardise two-factor authentication across all booking and ticketing platforms to keep your attendee data insulated from automated attacks.
Read more: bbc.co.uk