Dan

Embedded Security / Hardware Pentesting / LLM Tamer

[Dan @DanHardwareLab]

@elonmusk We accept the reality of the world with which we are presented. 🤓

[Dan @DanHardwareLab]

@natashenka Go back to being a fantastic IC

[Dan @DanHardwareLab]

@TracketPacer “the API bit over there”

[Dan @DanHardwareLab]

@lydiahallie Make being on CVP worthwhile. I have stopped using Fable 5

[Dan @DanHardwareLab]

@karpathy @DavidSHolz Does it view compacting its context as a threat? 😅

[Dan @DanHardwareLab]

@dhakal_ananda And I’m excited 👌👏👏👏👏👏

[Dan @DanHardwareLab]

@SciTechgovuk @biztradegovuk @DCMS @cabinetofficeuk @ProfBrianCox

[Dan @DanHardwareLab]

@karpathy This I found amusing. It’s why I prefer long incoherent rambles.. 😅

[Dan @DanHardwareLab]

@mdisec Cost that much in train fares 😅

[Dan @DanHardwareLab]

@jun_song Go back to 5.5?

[Dan @DanHardwareLab]

@RachelReevesMP “The economy is stronger”

Partly true, but hardly a triumph.😅

UK GDP grew by 1.3 per cent in 2025, and by 0.6 per cent in the first quarter of 2026. GDP per person was also 0.7 per cent higher than a year earlier.

[Dan @DanHardwareLab]

@stalman Localhost

[Dan @DanHardwareLab]

@JamesMelville BBC IN UHD

[Dan @DanHardwareLab]

I actually think the UK should spend MORE on fundamental research, not less. But we should also make our internal market much better at scaling companies, attracting investment and turning groundbreaking discoveries into world leading businesses. We shouldn't have to choose between great science and great commercialisation (IMHO 🤓)

[Dan @DanHardwareLab]

@ProfBrianCox A GPU accidentally producing a quantum field theory dad joke is peak LLM.

[Dan @DanHardwareLab]

@rickygervais I asked ChatGPT

[Dan @DanHardwareLab]

@ClarksonsFarm1 On that bombshell!

[Dan @DanHardwareLab]

@karpathy @btaylor Soon the ultiamte proof of humanity willbe three typos, a mising comma, and “sent from my iPhone”.

sent from my iPhone

[Zhengyao Jiang @zhengyaojiang]

The first experimental evidence of recursive self-improvement (RSI).

Autoresearching the autoresearch agent for eight days.

The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)

[Dan @DanHardwareLab]

@sama @fbettag Let’s not forget….