It's genuinely over

I've written about this exact scenario in my last article

Creating a gap between perception and reality is the literal WORST CASE for AI

If no one knows how capable the models are then no one is prepared.

you can read it here:

this seems like a terrible development

CAISI has reportedly been directed to stop publishing public model assessments as the new AI EO gets implemented.

Natsec engagement on AI is essential. But pulling CAISI's evals from public view doesn't make the field more secure. It just means fewer eyes on the science when we need more.

Openness and natsec don't have to be in tension here. We should be doing both.

Reporting by @AmrithRamkumar: https://www.wsj.com/politics/policy/white-house-reins-in-ai-testing-unit-as-national-security-concerns-grow-8bd33fbb?st=UpJxDg

@janet_e_egan You damn clowns @AnthropicAI

Meanwhile, the UK is lapping us in state capacity - just putting out good analysis alongside many major model releases with minimal drama

Also the UK government is also in constant turmoil but somehow AISI gets by. Can't the US make a carveout here too??

@tszzl I can't even understand why, exactly. If they need some top secret evals (plausible), just don't publish those.

@tszzl uh oh

@scaling01 Yup safety theater makes the threat invisible as we no longer know what they can do and bad actors only need to extract limited information they can do via distributed partial extractions where no one account fully violates rules or threat limits

@tszzl @bioshok3 The “evil ostrich” timeline has one of the highest % of coming to fruition - and it will get higher as desperation and opposition increase.

@scaling01 SBF literally attempted to gain control of the treasury dept while being a key investor in anthropic -- they(Effective Altruist's) literally attempted to make a full court play on the USG and don't forget the EA backed coup attempt with Helen tonner et al.

@tszzl looks like the universe is forcing yall to be the torch bearer and live up to the name of the company.

@losslandscape @tszzl white house wants messaging control and not to be undercut by CAISI saying x y and z are safe if they have separate cyber concerns

(this is bad to be clear but thats the thought process)

@tszzl sticks and stones...

@tszzl “Stop tracking jobs statistics and cost of living”

@janet_e_egan Beginning of the end of civilization. It was a good run

@tszzl quick

@janet_e_egan What's their reasoning for not publishing these publicly? Do you think uk aisi will do the same?

@tszzl @bioshok3 Probably on the advice of Sam Altman.