Matteo Prandi

Tech Lead AI Safety @ DEXAI - CTO @icarolab_ai | ISO/IEC Technical Expert on AI | Building evaluation frameworks to decode AI’s impact on society

[Arena.ai @arena]

Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5.

This is a 17-place jump from Kimi-k2.6 (#18 -> #1).

In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5.

The full model weights will be released by July 27.

Congrats to the @Kimi_Moonshot team on this major milestone!

[Matteo Prandi @emmepra]

@thsottiaux Multiple tabs in the app would make terminal close to useless

[Matteo Prandi @emmepra]

That is why EU should care less about prestige model races and more about AI Factories, data-centre capacity, and long-term access to frontier compute.

The outage lasted a few hours. The more serious risk is that one day the problem will not be downtime, but scarcity.

[Matteo Prandi @emmepra]

What struck me more than Claude going down last Monday was that almost nobody seemed to talk about the part that mattered most.

[Matteo Prandi @emmepra]

From a EU perspective, I do not think the answer is simply to build another model and hope it wins. The more important goal is assured, future-proof access to compute for startups, researchers, and public institutions if global access tightens.

[Sam Altman @sama]

Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.

In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome.

AI safety and wide distribution of benefits are the core of our mission. Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems. The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement.

We also will build technical safeguards to ensure our models behave as they should, which the DoW also wanted. We will deploy FDEs to help with our models and to ensure their safety, we will deploy on cloud networks only.

We are asking the DoW to offer these same terms to all AI companies, which in our opinion we think everyone should be willing to accept. We have expressed our strong desire to see things de-escalate away from legal and governmental actions and towards reasonable agreements.

We remain committed to serve all of humanity as best we can. The world is a complicated, messy, and sometimes dangerous place.

[Anthropic @AnthropicAI]

We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax.

These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.

[Matteo Prandi @emmepra]

@omarsar0 definitely on the same page here, difference is likely UX and userbase

[Matteo Prandi @emmepra]

@iyoushetwt SublimeText

[Matteo Prandi @emmepra]

@KiptheAI @arscontexta @getanima love it, 200k input tokens at every launch. amazing

[Matteo Prandi @emmepra]

@cgtwts unpopular unpopular opinion: SublimeText is better than Cursor

[Matteo Prandi @emmepra]

The visible glitch was trivial; the harder question is what kinds of instability remain invisible.

This keeps argumentative progression but avoids the “broken sentence per line” aesthetic that often makes threads look inflated.

[Matteo Prandi @emmepra]

I had a strange moment while coding with an AI agent this week, running on Kimi K2.5.

It wrote a closing HTML tag as </div㺞 with no apparent reason.

1/N

[Matteo Prandi @emmepra]

A rare ideograph stands out, but many deviations would not. They would be syntactically valid and semantically plausible.

As systems become more agentic, we shift from step-by-step controllers to designers of constraints, guardrails, and monitoring layers.