Tech
Front pageAI & LLMs
Revolut moves its developers to open AI models on its own servers and now pays a tenth
The British fintech has largely dropped paid services such as Claude Code for its coding assistant and now runs Chinese open models on its own Nvidia chips.
2 sources · The Register, StartupHub.aiAnthropic tests cryptographic proof that an answer really comes from the stated AI model
With the open research prototype FlashProve, users should be able to check that a provider is not quietly using a different model, without the model weights being made public.
2 sources · TechCrunch, Interesting EngineeringReuters: OpenAI staff were raising the alarm over internal security even before ChatGPT
The company disputes that it neglected security and announces a new internal security push together with the startup Cathedral.
2 sources · Reuters, Unite.AIProgramming languages
Twelve JavaScript frameworks, one task for an AI agent: Angular proves the most expensive, htmx and Astro the leanest
Developer Nolan Lawson had Claude Code build the same Pokémon web app each time and measured the cost in tokens, time and performance.
2 sources · The Register, Nolan LawsonAttacker could view secret settings of some customers of developer platform Trigger.dev
The TypeScript platform for AI agents and background jobs says the hole has been closed but is silent on the cause and the number of customers affected.
3 sources · Trigger.dev, The Register, RuntimeWireOther tech
New data thief has a language model estimate how much each victim is worth
Cisco Talos dissected VOLTA, Windows malware that sends stolen data to an AI model and switches to another model if it refuses.
2 sources · Unite.AI, AI Chat DailyPassword vault mSecure reports suspicious activity and leaves customers in the dark about what happened
After 'suspicious activity', users must renew passwords, two-factor authentication and passkeys, while it remains unclear whether their vault was accessed.
2 sources · The Register, Cyber Security NewsEighty serious incidents in a year, and London fears AI-assisted ransomware against vital services within two years
The NCSC counted 80 nationally significant incidents over the past year and is earmarking extra money for defence using AI agents.
2 sources · The Register, Cyber Security News