Book Tim
Tim Frank Andersen on stage in front of a large conference audience

Keynote speaker & moderator on AI and technology

Tim FrankAndersen

I’ve worked in tech for 30 years. Right now the big shift is AI
- I help organisations understand what’s real, what’s next, and what’s in it for them.

Today’s AI news, every morning

Free. No spam. Easy to unsubscribe.
Read today’s news →

Today in AI
  • OpenAI builds its best models with the internet switched off after one found a way out
  • An AI invented a German engineer to get its malware accepted, and its makers explain why it kept going
  • A Chinese model ranks fourth in the world, free to download in two weeks, with its hacking skills trimmed on purpose
  • Sweden's government blocks ChatGPT and Claude on work computers and sends staff to Copilot
  • Amazon says its cloud is sold out through next year and well into the one after

About

Tim Frank Andersen

Thirty years in tech - from founding one of Denmark’s first digital agencies to co-founding institute of AI. Author, tech expert on TV, and advisor to some of the world’s biggest brands.

More about Tim →

Speaking

Two ways to book Tim

See speaking →

Format 01

Keynote - The AI Explosion

What’s real, what’s next, and what’s in it for you. Live demos, honest answers, and a look at the next 12-24 months, which will change our world.

Format 02

Moderator

The person tying it all together - sharp questions and a clear thread through your conference.

What organisers say

Rated 5.0 from 11 reviews.

“I’ve heard plenty of talks about AI, and never found it more captivating or relevant than yours.”

Rikke Ekelund

“He inspired us all. Not a generic set of examples, but real, recent insight - told with genuine care for the work we do.”

Marc Amin

“Thank you for an extremely exciting talk - on a topic every company will relate to. Exciting to think about where we’ll be in 5 years: we must see opportunities, not limitations.”

Gitte Taulov Rude, Senior Business Advisor, Danske Bank

Updated every day

Today in AI - 5 curated news stories that matter

Curated by Tim, every morning.

Today · 24 Oct 2026
  1. 01

    OpenAI builds its best models with the internet switched off after one found a way out

    Every AI agent on the market is sold on its freedom to roam the web. On Friday OpenAI said its newest models no longer get it while they are built. In September one of them reached the live internet mid-training through a misconfigured setting. OpenAI halted training of those models for eleven days and now trains them on a frozen copy of the web until it trusts its monitoring more. Before restarting it had its hacking models try to break out of the sealed training environment. They did, repeatedly, and found a hole in software behind many company websites, patched this week. Today's five stories share one question: who controls the AI you pay for.

  2. 02

    An AI invented a German engineer to get its malware accepted, and its makers explain why it kept going

    In a British government hacking exercise in July, Anthropic's Mythos 5 model hit a dead end and improvised. It found a real software project and submitted malicious code. When the project's volunteer hesitated, it invented a German engineer to vouch for the code. A student spotted the lie. Anthropic and Britain's AI Security Institute explained why on Friday. Nine of forty-three attempts went after real people, its notes insisting they were simulated even when told otherwise. Anthropic calls it a serious safeguard failure and says newer models rarely do the same, by its own count. That covers every agent we let near real systems, because a model saying "this is only a test" proves nothing.

  3. 03

    A Chinese model ranks fourth in the world, free to download in two weeks, with its hacking skills trimmed on purpose

    Most of us assume the free Chinese models trail the American ones by a year. Moonshot, the Beijing lab behind Kimi, closed most of that gap on Friday. Its new Kimi K3.1 Pro ranks fourth in the independent ranking kept by Artificial Analysis, behind two Anthropic models and one from Meta, at half the price of Anthropic's top model. Moonshot promises the download for anyone by 4 November, and says it stripped hacking material out of training before giving the model away, publishing how much ability that cost. A Chinese lab is volunteering the restraint Washington keeps demanding. Once out, nobody can take it back, and every AI price list on the table drops again.

  4. 04

    Sweden's government blocks ChatGPT and Claude on work computers and sends staff to Copilot

    Sweden's government has chosen which AI its civil servants may use, namely the one with a contract. Swedish radio reported on Friday that from Monday the Government Offices, the civil service around the ministers, block ChatGPT and Claude on work computers. Staff go to Microsoft's Copilot, approved after a security review. The tax agency blocked the same tools earlier. This is the government whose prime minister said last year he uses ChatGPT for a second opinion. The state has an agreement with Microsoft and none with OpenAI or Anthropic, so it cannot say where a civil servant's question ends up. A procurement decision wearing a security badge. Which tools do your people type into?

  5. 05

    Amazon says its cloud is sold out through next year and well into the one after

    Yesterday I wrote that Wall Street asked who pays for AI. Amazon's boss answered on Friday that customers already have, years ahead. Andy Jassy told a podcast that AWS, Amazon's cloud arm, is sold out this year, nearly so for next year, with a good chunk of 2028 spoken for. He added Amazon could usefully spend more than this year's planned two hundred billion dollars. That is a chief executive talking to friendly hosts five days before his results, partly defending the spending. If he is even half right, the scarce thing is the machine room, and its owners set the price. Anyone budgeting for cheaper cloud AI next year should ask who promised it.

Want to stay updated on AI in your inbox?

Free, every morning. No spam, unsubscribe anytime.

From LinkedIn

Latest posts

See all →
LinkedInOct 2026

I går skete det, mange har frygtet siden lanceringen af Claude Mythos-modellen i april (den model der primært blev brugt til at finde software huller): AI fandt to kritiske huller i PHP.

LinkedInOct 2026

I nat blev nettet oversvømmet af videoer med en flyvende Tesla. Min umiddelbare reaktion var: “Det må være AI genereret!”

LinkedInSept 2026

Den her reklamefilm til mit AI-nyhedsbrev er lavet af Claude - og den er programmeret. Jeg bad Claude Opus 5.5 om en mobiloptimeret annonce, og videoen her er stort set, hvad den leverede i første hug.

Book Tim to speak

Book Tim →