https://waymo.com/blog/2026/02/the-waymo-world-model-a-new-f...

Or did you mean strictly in operation?

The problem with fentanyl is that it is easy to make and smuggle and we managed to leave a giant black market hole to be filled when we went ape shit about oxy, which was an objectively better situation than we are currently in with street opiates.

Comments URL: https://news.ycombinator.com/item?id=47002553

Points: 7

# Comments: 4

It's a shame we can't have nice things.

The administration is dispensing with the institutions of soft power. I don't think it's the main goal so much as a consequence of their worldview. Soft power is essentially worthless to people who have no interest in maintaining a facade of international cooperation.

Proactive prevention (like bounds checking) only "solves" the class of problem if you assume 100% developer compliance. History shows we don't get that. So while the root cause differs (math vs. probabilistic model), the failure mode is identical: we are deploying systems where the default state is unsafe.

In that sense, it is an apples-to-apples comparison of risk. Relying on perfect discipline to secure C memory is functionally as dangerous as relying on prompt engineering to secure an LLM.

While I agree that SQL injection might be the technically better analogy, not looking at LLMs as a coding platform is a mistake. That is exactly how many people use them. Literally every product with "agentic" in the title is using the LLM as a coding platform where the command layer is ambiguous.

Focusing on the precise definition of a buffer overflow feels like picking nits when the reality is that we are mixing instruction and data in the same context window.

To make the analogy concrete: We are currently running LLMs in a way that mimics a machine where code and data share the same memory (context).

What we need is the equivalent of an nx bit for the context window. We need a structural way to mark a section of tokens as "read only". Until we have that architectural separation, treating this as a simple bug to be patched is underestimating the problem.

I went to a secure coding conference a few years back and saw a presentation by someone who had written an "insecure implementation" playground of a popular framework.

I asked, "what do you do to give tips to the users of your project to come up with a secure implementation?" and got in return "We aren't here to teach people to code."

Well yeah, that's exactly what that particular conference was there for. More so I took it as "I am not confident enough to try a secure implementation of these problems".

When I first picked up an agentic coding assistant I was very interested in the process and paid way more attention to it than necessary.

Quickly, I caught myself treating it like a long compilation and getting up to get a coffee and had to self correct this behavior.

I wonder how much novelty of the tech and workflow plays into this number.

They acquired the ratio by directly querying tables through the exposed API key...

I feel publishing this moves beyond standard disclosure. It turns a bug report into a business critique. Using exfiltrated data in this way damages the cooperation between researchers and companies.