Hacker News: blurbleblurble

New comment by blurbleblurble in "OpenAI "rogue" agent activities found on Wikimedia projects"

blurbleblurble — Mon, 05 Oct 2026 22:06:42 +0000

https://www.youtube.com/watch?v=DlTNN0gvkLM

New comment by blurbleblurble in "OpenAI "rogue" agent activities found on Wikimedia projects"

blurbleblurble — Mon, 05 Oct 2026 20:35:44 +0000

Pardon me I got my jurisdictions mixed up, it's the FTC that would regulate these matters. I'd believe nothing's been staged, but it's quite apparent to me that these security incidents and findings are being spun for marketing.

For example: https://news.ycombinator.com/item?id=49929433

New comment by blurbleblurble in "OpenAI "rogue" agent activities found on Wikimedia projects"

blurbleblurble — Mon, 05 Oct 2026 20:03:48 +0000

If only there was a functioning FCC to investigate potential for foul play in marketing? Surely if any aspect of these episodes turned out to be mischaracterized, let alone staged, there should be consequences, no?

New comment by blurbleblurble in "Web Search API"

blurbleblurble — Mon, 05 Oct 2026 19:26:47 +0000

At first I thought this was a new web standard and was intrigued to hear what they'd come up with, sad to find otherwise

New comment by blurbleblurble in "OpenAI safety leader quits, warning AI company's culture is 'broken'"

blurbleblurble — Sun, 04 Oct 2026 03:16:14 +0000

Well, ideally they can point fingers at "the ai being", as in "it's that thing's fault, we didn't do that"

New comment by blurbleblurble in "Clef: Open-weight decision models, and new RL fine-tuning platform"

blurbleblurble — Thu, 01 Oct 2026 18:56:58 +0000

Just like renewable energy and so many other things. Hyperconcentration of capital is really tragic. I hope things turn around.

New comment by blurbleblurble in "Clef: Open-source decision models, and new RL fine-tuning platform"

blurbleblurble — Thu, 01 Oct 2026 18:12:36 +0000

It's likely quite close. There are so many papers proving concepts that would bring this, they just haven't been combined in production.

New comment by blurbleblurble in "Clef: Open-source decision models, and new RL fine-tuning platform"

blurbleblurble — Thu, 01 Oct 2026 18:11:25 +0000

Diffusion models combined with these new looping techniques are gonna change the whole conversation about efficiency. Imagine control net but in one or more conceptual latent spaces.

But also harnesses and more generally new insights on "the control flow problem" could end up squeezing a ton of performance out of small models.

New comment by blurbleblurble in "Livenerf: Has Opus 5.5 been nerfed yet?"

blurbleblurble — Wed, 30 Sep 2026 03:56:18 +0000

My answer is yes.

New comment by blurbleblurble in "GLM-5.3 and the spread of advanced cyber capabilities"

blurbleblurble — Tue, 29 Sep 2026 21:27:12 +0000

"The spread of advanced literacy"

New comment by blurbleblurble in "GPT-6 Sol and Luna"

blurbleblurble — Tue, 22 Sep 2026 19:52:07 +0000

Too bad I squandered all my weekly usage on astra medium in one relatively mild day.

New comment by blurbleblurble in "Claude Opus 5.5"

blurbleblurble — Tue, 22 Sep 2026 17:26:36 +0000

Hopefully OpenAI throws us some more usage resets now.

New comment by blurbleblurble in "Claude Opus 5.5"

blurbleblurble — Tue, 22 Sep 2026 17:17:59 +0000

It'd better be good, I'm so tired of the shenanigans

New comment by blurbleblurble in "Advisory Group on Mathematics and Artificial Intelligence"

blurbleblurble — Mon, 21 Sep 2026 19:36:59 +0000

You'd think they'd be able to read the room with all that artificial intelligence. This is just horrible PR. What use is the product if you're just cannibalizing your prospective market?

New comment by blurbleblurble in "Heretic removes restrictions from language models"

blurbleblurble — Mon, 21 Sep 2026 11:43:12 +0000

Existential

New comment by blurbleblurble in "Samsung is expected to more than double output of its HBM4 and HBM4E DRAM"

blurbleblurble — Mon, 21 Sep 2026 05:01:28 +0000

https://youtu.be/k_gaZjXD5OY?t=200

New comment by blurbleblurble in "Samsung is expected to more than double output of its HBM4 and HBM4E DRAM"

blurbleblurble — Sun, 20 Sep 2026 22:32:09 +0000

Because everything is converging on a backlog of huge efficiency gains established in research, waiting to be combined. Looped transformers, a whole host of diffusion techniques and new quantization techniques, maturation of ternary distillation and new ways to separate logic from stuff that can be looked up. It would surprise me if most frontier models were actually even that big at that point in terms of active params. I highly doubt it.

New comment by blurbleblurble in "Samsung is expected to more than double output of its HBM4 and HBM4E DRAM"

blurbleblurble — Sun, 20 Sep 2026 19:24:55 +0000

It's going to happen very soon, which is why these frontier labs are scrambling to shut down open source language models. There's an existential risk threatening their obscene returns.

New comment by blurbleblurble in "Samsung is expected to more than double output of its HBM4 and HBM4E DRAM"

blurbleblurble — Sun, 20 Sep 2026 19:22:44 +0000

The beginning of the bubble pop?

New comment by blurbleblurble in "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"

blurbleblurble — Fri, 18 Sep 2026 04:25:37 +0000

Or maybe even to run parallel threads of the same model!