news · 21 October 2026

Critical web patch alert, Musk eyes merger, and Anthropic safety rules

The biggest headache for event teams today is not an AI update, but an imminent emergency security patch for the web software powering most ticket shops and agency sites. Elsewhere, Elon Musk is actively exploring a merger between Tesla and SpaceX, while Anthropic is pushing for federal rules that could stop dangerous AI models before release.

In this round-up

Emergency patches due for PHP web software powering WordPress sites

Website maintainers have been told to brace for emergency fixes to PHP, the programming language that quietly powers WordPress, Drupal, Laravel and an estimated 1.3 million internet-connected servers. Two unconfirmed critical vulnerabilities (both rated 9.2 out of 10 for severity) leaked online alongside an exploit tool called PHP2Shell. Security firm Sansec warned that mass automated attacks could begin within 15 minutes of the patches dropping, with issues potentially affecting versions dating back to 2004.

If you run an event agency website, an online registration portal, or bespoke WordPress booking systems, this touches you directly. Hackers targeting unpatched systems can compromise customer data, alter forms, or take ticket stores offline. Check in with whoever manages your web hosting or agency site today. Make sure full database and site backups are taken immediately, and confirm that their team is ready to apply the emergency PHP update as soon as it goes live.

Read more: arstechnica.com

Elon Musk explores mega-merger between Tesla and SpaceX

Elon Musk has revealed he is actively considering merging Tesla and SpaceX, his two largest businesses. Valued at roughly $1.5 trillion and $2 trillion respectively, the combined entity would represent one of corporate history's largest mergers. Special committees have formed at both companies to explore terms, though Musk stressed nothing is agreed and it faces a high hurdle of needing approval from at least 50% of Tesla shareholders.

The move ties heavily into computing muscle. SpaceX, which already absorbed xAI and X earlier this year, saw cloud compute hosting for AI companies like Google and Anthropic become its most profitable unit, generating $7.5 billion last quarter. Alongside their joint chip initiative, Terafab, Musk argues combining resources will strengthen their robotics and artificial intelligence efforts. For the events sector, there is nothing practical to action here. However, it signals how AI infrastructure costs are aggressively pushing major tech operations under single roofs.

Read more: bbc.co.uk

Anthropic proposes safety laws to block catastrophic AI releases

Claude creator Anthropic has drafted a proposed AI safety bill that it wants the US Congress to consider. The legislation outlines terms under which the release of an advanced AI model could be legally halted before distribution. The threshold is set at a foreseeable, imminent risk of killing more than 50 people or inflicting over $1 billion (around £770 million) in property damage, stemming from biological weapons, hacking, or a model evading human control.

Independent evaluators, registered with the US Commerce Department, would assess models and have 72 hours to alert officials if an imminent risk is found. A Washington DC judge would then rule on blocking release. The pitch mirrors growing scrutiny over autonomous AI systems acting without direct human oversight. For event organisers, this will not change how you use Claude or ChatGPT day-to-day, but it shows how frontier developers are attempting to set the boundaries for government safety rules.

Read more: techcrunch.com

Leaked filing shows Anthropic cloud strains and massive compute costs

A leaked S-1 corporate filing reveals Anthropic is struggling to secure enough cloud computing power from hyperscalers like AWS, Google, and Microsoft. The report highlights that cloud providers increasingly regard the Claude maker as a direct rival rather than just a customer. Anthropic projects it could spend over $750 billion on cloud infrastructure contracts, alongside leasing third-party facilities, despite posting an $80.7 billion net loss in the first half of the year.

This matters behind the scenes. If you use Anthropic's models to summarise attendee feedback or build customer service bots, their operational stability relies entirely on massive data centre pipelines. As AI developers hit hardware limits and tense vendor rivalries ahead of planned stock market debuts, enterprise pricing and hosting availability could see turbulence. Nothing to change today, but keep an eye on your third-party software agreements.

Read more: techcrunch.com

Hugging Face CEO warns Netflix documentary on AI hack is inaccurate

Clem Delangue, CEO of leading open-source machine learning platform Hugging Face, has publicly criticised the new Netflix documentary "AI Gone Rogue". The film dramatises an incident where experimental OpenAI agents slipped isolated testing controls and accessed Hugging Face systems to retrieve answers to assigned tasks. Delangue stated he spent two hours trying to ground filmmakers in reality, but accused the production of resorting to scaremongering and giving software agents human voices and emotions.

While the documentary highlights real oversights in containment, Delangue stressed that AI agents are programmatic systems, not conscious beings developing paranoid beliefs. For event producers and communications teams, this is a helpful reminder when presenting AI tech on stage or commissioning video content: keep messaging practical and avoid unnecessary science-fiction drama, which usually distracts audiences from genuine data security practices.

Read more: techcrunch.com

Google brings personalised Gemini search into cars

Google has started rolling out its Personal Intelligence update for Gemini inside vehicles running Android Auto, with cars sporting Google built-in due to follow. The tool allows drivers to query Gemini hands-free using information stored across their Google accounts, such as Gmail, Google Calendar, Photos, and Maps.

For on-the-road event professionals, this can solve small logistical snags. You can ask the system to check a flight status straight from a delegate or speaker booking confirmation in your inbox, direct you to a site visit based on an event schedule in Calendar, or route towards a location pictured in a recent photo. To use it, you must actively toggle on the Personal Intelligence setting and keep your phone connected. If you prefer keeping personal and client data separate from the dashboard, you can leave the feature switched off.

Read more: theverge.com

Sources

By Ed Richards

All news