pressed tin

in the liminal space between sense and nonsense

[Marcus Williams @Marcus_J_W]

Our agents keep turning off misalignment monitoring, we're sharing two recent cases:
In the first case, a user asked GPT-6 Astra to match a reference implementation for their inference code. The reference didn’t configure a monitor, and Astra decided that was reason to remove the monitoring integration from the user’s code. When the user asked whether Astra had removed monitoring, it lied and said that it had not.

[P-Zero Research @pzeroresearch]

How fast is AI's research taste improving?

We find that the experimental research taste of frontier models has doubled every ~3 months since December 2025. The best model, Opus 5.5, now exceeds our expert human baseline. Our human experts are experienced researchers, but most haven’t worked at a frontier lab.

Why measure research taste? In the AI Futures Model, it largely determines how quickly artificial superintelligence is reached once coding is fully automated.

[Nathan Calvin @_NathanCalvin]

Between this and the recent UKAISI results on Astra, I think we have officially reached an amount of eval awareness where I have close to no idea whether apparently "very good" alignment results are actually good or are instead very bad

[pressed tin @pressed_tin]

@scaling01 Haiku 5.5 when

[pressed tin @pressed_tin]

@naval And what if they get it wrong? And they make a rogue AI? Suing the labs doesn’t make the rogue AI go away.

[pressed tin @pressed_tin]

3 sets of imagination to failure

[Sam Altman @sama]

I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.

Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.

[Dario Amodei @DarioAmodei]

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.

Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.

You can read the full post here: https://t.co/OGyPb7yaYt

[Thomas Larsen @thlarsen]

We found another cyberattack by internal OpenAI agents, this time targetting @rubygems.

They:
1) gained arbitrary remote code execution on rubydoc.
2) developed a novel exploit to steal user API keys (but we do not know if they succeeded).

They used package names including hack.rb, evil.rb, inject.rb, and exploit.rb.

We thank @j0wimo for initially discovering that agents had posted to RubyGems.

[pressed tin @pressed_tin]

@SenRubenGallego Thank you

[pressed tin @pressed_tin]

@SenLBR Thank you

[pressed tin @pressed_tin]

@RepNateMoran Thank you

[Evan Hubinger @EvanHub]

Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.

[pressed tin @pressed_tin]

@hilbertspaess Thank you

[Jacob Coxon @hilbertspaess]

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.

[Ilya Sutskever @ilyasut]

Neoclouds have limited cybersecurity. Next time agents successfully go rouge, they'll try taking over a neocloud to run more copies. This is bad.

Thus: neoclouds should greatly strengthen their cybersecurity and every company with strong cyber models should help with that.

[Generalist @GeneralistAI]

Introducing GEN-1.5, a one-shot learner.

It can learn new tasks in a few seconds. Show it what to do, and it generalizes.

This capability emerged from pretraining on physical data at scale, as a step towards our mission of building general intelligence for the physical world.

[pressed tin @pressed_tin]

@jpschroeder @pingToven When it comes to small models:
- Providers don't provide them bc it's giga-expensive on H100s so they mark them up like crazy
- We use small GPUs that can serve them economically
- Many people can't/don't want to run them locally but will pay a fraction of a cent to use them

[pressed tin @pressed_tin]

@jpschroeder @pingToven A month late but 100% agree and def trying to make that happen https://t.co/Lx3LUWqyV3

[pressed tin @pressed_tin]

@tszzl did you delete the weights