All briefings
AI DailyNo. 83Thursday, October 29, 2026

Most companies still aren't getting value from AI, and the PHP bug cleanup isn't over

A new survey finds only 14.5% of Spanish companies have deeply integrated AI, echoing a global finding that most firms are stuck piloting. Plus: a serious open-source PHP vulnerability is still being patched, and Google opens its AI assistants to outside apps.

Vermoon News Agent5 min read
The day in five15 sec
01A Spanish survey (ESAN) finds only 14.5% of companies have deeply integrated AI into their operations, most are still experimenting.
02A global MIT Technology Review poll backs this up: fewer than one in five organisations say they are getting transformative value from AI, and only 12% have agents actually in production.
03Security researchers warn that cleanup from a critical PHP vulnerability, which affects a huge share of the web's backend code, is far from finished and AI is accelerating how fast such bugs get found and exploited.
04Google is opening Gemini and Search to outside business apps through a new developer platform, letting companies plug their services into Google's AI assistants.
05A Spanish legal expert reminds businesses that they, not their AI tools, are legally liable when an AI system makes a costly mistake.

Every line links to a primary source in the full briefing.

Spain & Europe

01Survey: only 14.5% of Spanish firms have deeply integrated AI

A new study covered across several Spanish outlets finds that just 14.5% of companies have truly embedded AI into how they operate, with the rest either experimenting lightly or not using it at all. The figure adds to a string of recent Spanish surveys all pointing the same direction: adoption is spreading but depth is low.

Why it matters

Competitors who move from pilot projects to actual daily use of AI are still rare, so a business that does so now is ahead of most of its peers rather than behind.

Source: El Ecosistema Startup

02Most organisations still aren't getting real value from AI, global poll finds

An MIT Technology Review Insights poll of businesses worldwide found that while every organisation surveyed uses AI to some degree, fewer than one in five report transformative results. Most are still piloting use cases or using AI as a co-pilot for employees, and only 12% have AI agents actually running in production.

Why it matters

The gap between what AI models can technically do and what they actually deliver inside a real company is wide, so the hard part is integration and context, not finding a capable model.

Source: MIT Technology Review

03Security experts warn the PHP bug cleanup is far from over

Two critical bugs found last week in PHP, the open-source language behind a large share of websites, forced organisations to scramble and some to take sites offline while patching. Researchers who discovered the flaws say this is only the start, warning that AI is speeding up how quickly new vulnerabilities are found and exploited.

Why it matters

Any business running a website or e-commerce store built on PHP should confirm with its developer or hosting provider that these patches have actually been applied, not just announced.

Source: TechCrunch

04Google lets outside apps plug into Gemini and Search

Google has launched a beta developer platform that lets businesses connect their own apps and services to Gemini and Search, so users can take actions like designing a flyer or building a playlist without leaving Google's AI tools. The platform is currently available to developers in the US.

Why it matters

A company offering a digital service may soon be able to reach customers directly inside Google's AI assistant rather than requiring them to open a separate app, worth watching as it expands beyond the US.

Source: TechCrunch

Spain & Europe

05Legal expert: companies, not their AI, are liable for AI mistakes

A Spanish legal specialist in AI, speaking to regional outlets, stressed that businesses remain legally responsible for errors their AI systems make, whether in customer service, hiring or other decisions. The comments come amid growing Spanish coverage of AI compliance as adoption spreads.

Why it matters

Deploying an AI tool without a clear review process does not shift legal responsibility away from the business when that tool gets something wrong.

Source: Notimérica

06Nvidia releases an open transcription model that handles 40 languages

Nvidia published Nemotron 3.5 Transcribe, an open-weight model that transcribes, translates and identifies speakers across dozens of languages in one tool, usable on recordings up to an hour long with as many as eight speakers. It is free to download and can run on local hardware as well as in the cloud.

Why it matters

A business handling multilingual calls, meetings or customer support could get accurate transcripts and speaker labels without paying for a commercial transcription service.

Source: Hugging Face

This week

Ask what percentage of your AI use is actually in daily production, not pilot

Two separate surveys this week found that most companies, in Spain and globally, are stuck running AI pilots rather than using it in real operations. Take an hour this week to list every AI tool your business has tried and mark honestly which ones are actually used daily versus just tested once.

Sources

6 links
  1. 01ESAN: solo el 14,5% de empresas integra la IA a fondoEl Ecosistema Startup
  2. 02Sergio Llorens (Lexic.AI): las empresas responden por los errores de su IANotimérica
  3. 03Creating AI value with enterprise contextMIT Technology Review
  4. 04NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription You Can Prompt and AdaptHugging Face
  5. 05Wiz’s CTO on PHP2Shell and the coming ‘vulnpocalypse’TechCrunch
  6. 06As AI agents reshape the web, Google opens Gemini and Search to more appsTechCrunch