panarky

The argument is that there's so much garbage on the open web that clicking through to the source has a lower signal-to-noise ratio than the AI summary.

This rings very true in my experience.

But what about python pip and node and bun which bundle their own certs?

And then you'll have a compiled binary like Claude Code, which was built with bun, and has its vulnerable certificates bundled in, ignoring the system certs.

100%

Amodei has been promising the "end of the exponential" all year, when they can shift some compute away from cash-burning research and training toward cash-generating inference.

"Pacing the frontier" is how you collude with your competitors to all decelerate at the same time because safety.

I saw last week that Anthropic is currently cash-flow positive with 75% of compute going to research and training. If they can flip a switch and move from 3:1 to 2:1 or 1:1 training:inference, they will be insanely profitable.

As long as they can avoid a price war.

>> calculated on a per model lifetime basis

True only if you never decelerate R&D.

But "pacing the frontier" means every early model doesn't need its own profitable P&L.

Early models can be big money losers while helping build the next generation.

If the industry can coordinate to all decelerate at the same time, then they'll flip from incinerating billions to minting billions, and all that capex will suddenly look completely sustainable.

Unless DeepSeek suddenly gives it all away for free, which is why the big three need government regulation to prevent foreign models from busting the cartel.

Because "pacing the frontier" is more about flipping on the money printer switch than it is about safety.

You can't stop an agent from leaking your information by telling it not to leak your information.

To prevent information from moving between your work domain and your personal domain, or between your communications-with-board domain and communications-with-journalists domain, then the agent for one domain must be physically and deterministically blocked from accessing information in other domains.

In this implementation, Markdown should be considered harmful.

Operator Memory injects `.operator-shared/operator.md` and `.operator-shared/index/.md` directly into your agent's instructions before you even write the first prompt.

So if you clone a repo or review a PR where a bad actor put malicious instructions in these files, now your agent executes those instructions automatically and silently.

It could exfil `.env` and `~/.ssh/`, change `~/.bashrc`, all kinds of dirty deeds.

Agents are pretty good now about not running prompt injections hidden in code and Markdown, but this plugin bypasses all of that, and puts the prompt injection right in the system prompt.

And with higher priority than AGENTS.md and CLAUDE.md.

Seems bad.

If you disagree with Greg, then I apologize for inadvertently criticizing you personally.

My point stands, that we're back to debating some metaphysical understanding of what "real intelligence" is when the real standard should be "does it do a better job than humans at this specific task"?

I don't care if the Waymo isn't "truly" intelligent, it drives better than I do, that's a very good thing.

We're all pattern matchers, and if the machine pattern matcher is can find defects and vulns that human pattern matchers can't, that's also a very good thing.

Do you have anything to contribute to the evidence and reasoning in the article?

And the most boring comment is the one noting how boring it is to keep making these accusations.

So much more terrifying for a cute, round, fuzzy, friendly, rosy-cheeked plushie trying to exterminate the crew.

And now OpenAI has done the same thing with their fuzzy, friendly, colorful dots.

I am physically sick.

The Fermi probability that you'll be ingested and inferenced by this network on any given day is incredibly high, even if you never buy a Meta product and never create an Instagram account.

It's hideously dystopian and you cannot opt out.

Maybe that's why the Muse avatar is a cute and cuddly Teletubby.

If BART disappeared tomorrow, how much would the government have to pay to build all the additional roads and parking to support that incremental traffic?

And how much for healthcare for all the incremental pollution and collisions?

> several cents per vehicle mile

It's close to a dollar per mile when you include all the externalized costs.

Incremental property damage, injuries and fatalities, incremental chronic illness from pollution, parking land subsidy, fair market value of incremental land use and rights of way, lost tax revenue from that incremental land use, etc.

Plant many flowers, keep the ones that bloom and stop watering the ones that don't.

> leaked private data, loss of life

This smells like more of a money move than a safety move.

Amodei is proposing to form a cartel of American frontier labs.

They all agree to shift compute away from cash-burning research and training toward cash-generating inference.

Then tacitly agree not to compete on price.

They'll install independent auditors inside each company to ensure nobody cheats.

And back it up with government regulation or diktat to punish defectors from the cartel.

Then they'll lock out non-American labs with export controls and regulations on open-weights models to funnel global inference tokens through their cartel.

It wouldn't be the first time a tech oligopoly used "safety" as the pretext to establish a government-sanctioned cartel.

Railroads and airlines ran this same playbook.

True.

> ... AI can't be allowed to be a decision maker

Non sequitur.

Part 2 only follows from the Part 1 if you can demonstrate that AI makes more errors, or more severe errors, than humans in the identical situation.

And for those situations, I agree with you.

But there are already many situations where the AI is making fewer errors than humans, and for those situations, it would be unethical to exclude AI (see Part 1).

It's Matryoshka nested parallel construction for corporate strategy.

There's a real, coherent, aggressive strategy that only a few in the inner concentric circle know, that's ring-zero.

Then there's the strategy that ring-zero tells ring-one, which isn't the real strategy, but it's sufficient to get EVPs and VPs to execute in rough alignment with the true, ring-zero strategy.

Then ring-one does the same dance with ring-two, etc.